Artificial intelligence-based sepsis prediction system and method

Abstract

The present disclosure encompasses systems and methods for determining whether an individual is present with sepsis or determining the risk of developing a high risk individual to sepsis, including within about 24 hours after obtaining measurements and / or information of one or more factors associated with the individual. In particular embodiments, an output (e.g., a quantitative value) may be generated using one or more input characteristics (e.g., one or more demographic information, patient evaluation, vital signs, hematological laboratory measurements, chemical panel measurements, and / or sepsis-related biomarkers), the output indicates that the individual's risk of developing sepsis is a low risk, a medium risk, a high risk, or a very high risk.

Description

[0001] Cross-reference to related applications This application claims priority and interest in U.S. Provisional Patent Application No. 63 / 459,568, filed April 14, 2023, the entire contents of which are incorporated herein by reference. Technical Field

[0002] This disclosure generally relates to methods, compositions, and systems that are at least related to medicine, physiology, and biology. More specifically, this disclosure relates to risk assessment of sepsis attacks. Background Technology

[0003] In many diseases and conditions, positive outcomes of treatment and / or prevention are closely associated with early and / or accurate diagnosis. However, effective early diagnosis and treatment methods are often lacking, so treatment is often administered too late, inappropriately, or to individuals from whom it cannot benefit.

[0004] Sepsis can be a serious, often life-threatening illness requiring urgent and comprehensive treatment. Treatment depends on the type of infection, but antibiotics or similar medications are usually the first line of treatment. Because sepsis can be caused by a variety of microbial infections, the disease is difficult to predict and diagnose early, making effective intervention challenging. It is an excessive and uncontrolled inflammatory response in the body, often due to an inappropriate reaction of the individual's immune system to pathogens. Furthermore, the presence of large amounts of pathogens in palpable sites or bodily fluids in infected individuals further complicates diagnosis. Therefore, regardless of the cause, it is necessary to identify biomarkers indicating sepsis risk or early onset for early and effective intervention. Differentiating between patients at risk of developing sepsis and those not at risk also aids in managing the condition. Therefore, there is an urgent need to identify measurable, disease-specific biomarkers that indicate the risk of progression to sepsis or early onset, as well as methods for screening using these biomarkers. Summary of the Invention

[0005] Embodiments of this disclosure include a method for generating a sepsis risk indicator, the method comprising: receiving at least one input feature corresponding to a patient; analyzing the at least one input feature using a machine learning model to generate a predictive score based on the at least one input feature, the predictive score indicating the probability that the patient has or develops sepsis over a predetermined time span; wherein the at least one input feature is associated with sepsis.

[0006] In some embodiments, a method for providing a diagnosis of sepsis is provided, the method comprising: receiving patient-related data, the data containing concentrations of one or more input features; analyzing the data using at least one machine learning model to generate a predictive score based on at least one input feature selected from a set of input features listed in any one of Tables 8-13, the predictive score indicating the probability that the patient has or develops sepsis within a predetermined time span, wherein the sets of input features listed in Tables 8-13 are associated with sepsis, and wherein the sets of input features are listed in Tables 8-13 according to their relative importance to the predictive score; and generating a diagnostic output based on the predictive score. In a specific embodiment, generating the diagnostic output includes: determining that the predictive score belongs to one of four categories; and generating the diagnostic output based on the category to which the predictive score belongs, the diagnostic output including: the probability that the patient has or develops sepsis within the predetermined time span is low, moderate, high, or very high. In some embodiments, the predictive score ranges from 0 to 1. In some embodiments, the four categories include a low category, a medium category, a high category, and an extremely high category, wherein the low category is specified as the predicted score being below a first threshold, the medium category is specified as the predicted score being between the first threshold and a second threshold, the high category is specified as the predicted score being between the second threshold and a third threshold, and the extremely high category is specified as the predicted score being above the third threshold.

[0007] In a particular embodiment, there is a method for generating a sepsis diagnosis, the method comprising: receiving input features corresponding to a patient; and analyzing the input features using a machine learning model to generate a predictive score based on at least one input feature selected from any of the input feature groups listed in Tables 8-13, the predictive score indicating the probability that the patient has or develops sepsis over a predetermined time span, wherein the input feature groups listed in any of Tables 8-13 are associated with sepsis, and the input feature groups are listed in Tables 8-13 without a specific order of importance. Furthermore, the input feature groups listed in Tables 8-13 can be considered to be listed according to their relative importance to the predictive score. Any method can generate a diagnostic output based on the predictive score. In some embodiments, the input features include at least one clinical parameter and at least one biomarker listed in any of the input feature groups listed in Tables 8-13. Attached Figure Description

[0008] To gain a more comprehensive understanding of the principles and advantages disclosed herein, the following description will now be made with reference to the accompanying drawings, in which: Figure 1This describes the inputs, functions, artifacts, intermediate outputs, and final outputs used and generated in the sepsis immune scoring algorithm example. Functions are represented as trigger logic, imputing input features, generating a pre-calibration score function, generating a SHAP value, calibrating a prediction, and generating a risk category. Intermediate data is reflected as input features (unimputed) and the pre-calibration sepsis immune score. Artifacts are reflected as cubic steps. The final output is represented as input features, feature SHAP values, the sepsis immune score, and the sepsis immune score risk category.

[0009] Figure 2 These are example methods for generating sepsis risk indicators according to various embodiments.

[0010] Figure 3A An example of using the SHAP value of the first body of Model 1A is provided. Figure 3B An example of using the SHAP value of the second body of Model 1A is provided. Figure 3C An example using the SHAP values ​​of the third body from Model 1A is provided. In all cases, the values ​​that increase the risk of sepsis are represented by the bars on the right, while the values ​​that decrease the risk of sepsis are represented by the bars on the left.

[0011] Figure 4A An example using the SHAP value of the first body of Model 2 is provided. Figure 4B An example using the SHAP value of the second body in Model 2 is provided. Figure 4C An example using the SHAP values ​​of the third body from Model 2 is provided. In all cases, the values ​​that increase the risk of sepsis are represented by the bars on the right, while the values ​​that decrease the risk of sepsis are represented by the bars on the left.

[0012] Figure 5A An example using the SHAP value of the first body of Model 3 is provided. Figure 5B An example using the SHAP value of the second body in Model 3 is provided. Figure 5C An example using the SHAP values ​​of the third body from Model 3 is provided. In all cases, the values ​​that increase the risk of sepsis are represented by the bars on the right, while the values ​​that decrease the risk of sepsis are represented by the bars on the left.

[0013] Figure 6A An example using the SHAP value of the first body of Model 4 is provided. Figure 6B An example using the SHAP value of the second body in Model 4 is provided. Figure 6C An example using the SHAP values ​​of the third body from Model 4 is provided. In all cases, the values ​​that increase the risk of sepsis are represented by the bars on the right, while the values ​​that decrease the risk of sepsis are represented by the bars on the left.

[0014] Figure 7A An example using the SHAP value of the first body of Model 5 is provided. Figure 7B An example using the SHAP value of the second body in Model 5 is provided. Figure 7C An example using the SHAP values ​​of the third body from Model 5 is provided. In all cases, the values ​​that increase the risk of sepsis are represented by the bars on the right, while the values ​​that decrease the risk of sepsis are represented by the bars on the left.

[0015] Figure 8 It is a block diagram of a computer system according to various embodiments.

[0016] Figure 9A shows a non-limiting example of two sets or box features applied to Model 1A according to various embodiments.

[0017] Figure 9B shows a non-limiting example of three sets or box features applied to Model 1A according to various embodiments.

[0018] It should be understood that the accompanying drawings are not necessarily drawn to scale, and the proportional relationships between objects in the drawings are not necessarily to scale. The drawings are intended to provide a clear and understandable description of various embodiments of the apparatus, systems, and methods disclosed herein. Where possible, the same reference numerals will be used in all drawings to refer to the same or similar parts. Furthermore, it should be understood that the drawings are not intended to limit the scope of this disclosure in any way. Detailed Implementation

[0019] I. Exemplary Explanation of Terminology As used in the specification, "a" or "an" may refer to one or more. As used in one or more claims, when used in conjunction with the word "comprising," the word "a" or "an" may refer to one or more. Some embodiments of this disclosure may include or primarily include one or more elements, method steps, and / or methods of this disclosure. It is contemplated that any method or composition described herein can be implemented with respect to any other method or composition described herein, and different embodiments may be combined.

[0020] Unless explicitly stated otherwise, the use of the term "or" in the claims is intended to mean "and / or," although this disclosure supports the definition of "and / or" referring only to alternatives. For example, "x, y, and / or z" can refer to a single "x," a single "y," a single "z," "x, y, and z," "(x and y) or z," "x or (y and z)," or "x or y or z." Specific interpretations of x, y, or z may be specifically excluded from the embodiments. As used herein, "another" can refer to at least a second or more.

[0021] The term "several" means more than one.

[0022] As used in this article, the term "multiple" can mean 2, 3, 4, 5, 6, 7, 8, 9, 10 or more.

[0023] As used in this article, the term "set" refers to one or more items. For example, an itemset includes one or more items.

[0024] As used herein, the phrase “at least one of…” when used with a list of items means that different combinations of one or more items from the list may be used, and only one item from the list may be required. The item can be a specific object, item, step, operation, process, or category. In other words, “at least one of…” means that any combination or any number of items from the list may be used, but not all items in the list may be required. For example, but not limited to, “at least one of item A, item B, or item C” means item A; item A and item B; item B; item A, item B, and item C; item B and item C; or item A and item C. In some cases, “at least one of item A, item B, or item C” means, but not limited to, two of item A, one of item B, and ten of item C; four of item B and seven of item C; or some other suitable combination.

[0025] As used herein, “substantially” means sufficient to achieve the intended purpose. Therefore, the term “substantially” allows for minor, insignificant deviations from absolute or perfect conditions, dimensions, measurements, results, etc., such as those expected by a person of ordinary skill in the art that do not significantly affect overall performance. When used with respect to numerical values, parameters, or characteristics that can be expressed as numerical values, “substantially” means within ten percent.

[0026] In this specification, unless the context otherwise requires, the words “comprising,” “including,” and “having” will be understood to imply inclusion of the stated steps or elements or groups of steps or elements, but do not exclude any other steps or elements or groups of steps or elements. “Constitutes of” means including and limited to what follows the phrase “consisting of.” Therefore, the phrase “consisting of” indicates that the listed elements are necessary or mandatory, and no other elements may be present. “Mainly constitutes of” means including any elements listed following the phrase, and is limited to other elements that do not interfere with or contribute to the activities or actions specified for the listed elements in this disclosure. Therefore, “mainly constitutes of” indicates that the listed elements are necessary or mandatory, but no other elements are optional, and other elements may or may not be present depending on whether they affect the activities or actions of the listed elements.

[0027] References to "an embodiment," "an embodiment," "a specific embodiment," "a related embodiment," "a particular embodiment," "an additional embodiment," or "a further embodiment," or combinations thereof, in this specification refer to a specific feature, structure, or characteristic described in connection with that embodiment being included in at least one embodiment of the invention. Therefore, the appearance of these phrases throughout this specification does not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in various embodiments.

[0028] The term "artifact" as used in this article refers to the digital product used in the algorithm.

[0029] As used herein, "artificial neural network" or "neural network" (NN) can refer to a mathematical algorithm or computational model that simulates a set of interconnected artificial nodes or neurons that process information based on connectionist computational methods. A neural network (also called a neural mesh) may employ one or more nonlinear units to predict outputs in response to received inputs. Some neural networks include one or more hidden layers in addition to an output layer. The output of each hidden layer serves as the input to the next layer in the network (i.e., the next hidden layer or output layer). Each layer of the network generates an output from the received inputs based on the current values ​​of its corresponding set of parameters. In various embodiments, references to "neural network" may refer to one or more neural networks.

[0030] Neural networks can process information in two ways: in training mode when they are being trained, and in inference (or prediction) mode when they put what they have learned into practice. Neural networks learn through a feedback process (such as backpropagation), which allows the network to adjust the weights of individual nodes in the intermediate hidden layers (changing their behavior) so that the output matches the output of the training data. In other words, neural networks learn by being fed training data (learning examples) and eventually learn how to obtain the correct output, even when faced with a new range or set of inputs. Neural networks can include, but are not limited to, at least one of the following: feedforward neural networks (FNNs), recurrent neural networks (RNNs), modular neural networks (MNNs), convolutional neural networks (CNNs), residual neural networks (ResNets), neural networks for ordinary differential equations (neural-ODEs), or another type of neural network.

[0031] As used herein, the terms “adjudication,” “determination,” or “in the process of adjudication” refer to a physician adjudication process in which one or more physicians review a patient’s medical record, typically through a graphical review, to retrospectively determine what has happened. As used herein, the term “biomarker” generally refers to any measurable substance extracted from a subject as a sample whose presence indicates the presence of a phenomenon. Examples of such phenomena may include the subject’s age, disease state, condition, or exposure to a compound or environment. In the various embodiments described herein, biomarkers included in the analysis (also employing age-related biomarkers) may be used for diagnostic and / or therapeutic purposes (e.g., diagnosing a health condition or disease state). The term “biomarker” may be used interchangeably with the term “marker.”

[0032] As used herein, the term "patient" generally refers to a mammalian subject. A mammal can be a human, or an animal, including but not limited to horses, pigs, dogs, cats, ungulates, and primates. In one embodiment, the individual is a human. The methods and uses described herein are intended for medical and veterinary use. "Patient" is a human subject unless otherwise stated.

[0033] The term "training data" as used in this article generally refers to data that can be input into models, statistical models, algorithms, and any system or process that can make predictions using existing data.

[0034] As used in this article, a “model” can include one or more algorithms, one or more mathematical techniques, one or more machine learning algorithms, or a combination thereof.

[0035] As used in this article, "machine learning" can be the practice of using algorithms to analyze data, learn from it, and then make judgments or predictions about things in the real world. Machine learning algorithms can learn from data without relying on rule-based programming. Machine learning algorithms can include parametric models, nonparametric models, deep learning models, neural networks, linear discriminant analysis models, quadratic discriminant analysis models, support vector machines, random forest algorithms, nearest neighbor algorithms, combined discriminant analysis models, k-means clustering algorithms, supervised models, unsupervised models, logistic regression models, multiple regression models, penalized multiple regression models, or another type of model.

[0036] The term “LOINC” as used in this article refers to Logical Observation Identifier, Name, and Code (LOINC), a standard that facilitates the exchange and aggregation of outcomes (e.g., laboratory tests or vital signs) for use in clinical care, outcome management, and research.

[0037] In some embodiments, the term “sepsis immune score” as used herein refers to the probability that a patient will develop or progress to sepsis (as defined by sepsis-3) within 24 hours from the order time of the sepsis immune score.

[0038] The term “sepsis immune scoring algorithm” as used in this article refers to the process of generating a sepsis immune score using real-time measurements from a group of patients.

[0039] The term "SHAP value" used in this article refers to the Shapley value, which is a value associated with the contribution of each input feature to the final prediction.

[0040] II. Overview of Implementation Examples Embodiments of this disclosure include generating any type of indicator or output, such as quantitative output (e.g., a score) or qualitative output, reflecting an individual's risk of having sepsis or developing sepsis (e.g., risk relative to the general population), including at least some cases that developed sepsis within a specific time period. Sepsis can be of any type, including sepsis, severe sepsis, or septic shock. In specific embodiments, the output may or may not be transmitted in real time. In some embodiments, the individual has an infection defined by the presence of one of the following criteria: 1. Four days or more of qualified antibiotic use days (QAD) as defined by the Centers for Disease Control and Prevention (CDC).

[0041] 2. The microbiological test or culture result is positive, but does not include known contaminants (known contaminants in urine culture can be defined as "Lactobacillus, Corynebacterium, Gardnerella, α-hemolytic streptococci, and aerobic bacteria").

[0042] QAD guidelines follow the description of infection in the CDC Adult Sepsis Hospital Toolkit (CDC Adult Sepsis Hospital Toolkit https: / / www.cdc.gov / sepsis / pdfs / Sepsis-Surveillance-Toolkit-Mar-2018_508.pdf).

[0043] In certain embodiments, the methods and compositions of this disclosure relate to the probability or risk of a patient having or developing sepsis within 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24 hours or longer from the time of patient assessment, as defined by Sepsis-3 (Singer et al., JAMA. 23 Feb 2016; 315(8): 801–810, the entire contents of which are incorporated herein by reference). The output of the patient analysis may be referred to herein as the Sepsis ImmunoScore.

[0044] The output of the method can be a continuous value or a non-continuous value, such as a value between 0 and 100. In at least some cases, the output can be displayed to the user as software (SaMD) of a medical device. In some embodiments, the output can be interpreted as the probability that a patient has or has developed sepsis (as defined by Sepsis-3) within 24 hours from the date of the sepsis immune score prescription. In some embodiments, the sepsis immune score can be differentiated from an uncalibrated sepsis immune score prediction by applying calibration, which improves the interpretability of the uncalibrated sepsis immune score prediction.

[0045] In various embodiments, it is necessary to diagnose sepsis or determine the individual's risk of developing sepsis. The individual can be of any age, but in specific embodiments, the individual is at least about 65 years old or less than about 1 year old. The individual may be currently hospitalized or have recently been hospitalized. The individual may be immunocompromised or immunocompetent. In some embodiments, the individual may have been infected or is at risk of infection, including bacterial, viral, fungal, prion, or parasitic infections. The individual may or may not have chronic diseases such as heart disease, lung disease, cancer, kidney disease, hepatitis, and / or diabetes, as examples only. In specific embodiments, the individual may be infected with SARS-CoV-2, influenza, meningitis, pneumonia, tuberculosis, Escherichia coli, Clostridium difficile, etc.

[0046] Patients may experience one or more symptoms of sepsis, including rapid heart rate or weak pulse; fever, shivering or feeling very cold; confusion or disorientation; rapid breathing; severe pain or discomfort; and / or cold, clammy skin or sweating.

[0047] The individual may live temporarily or permanently in an institution with other people, such as a nursing home, rehabilitation center, professional care facility, hospital, etc.

[0048] In a specific embodiment, the individual is an adult aged 18 or older, and a serious infection is suspected based on the initial blood culture order in an emergency room or hospital setting.

[0049] In specific embodiments, the methods and systems of this disclosure employ one or more of the following: one or more demographic measurements, one or more patient assessments, one or more vital signs, one or more hematological laboratory values, one or more chemical laboratory values, and / or one or more sepsis-related biomarker concentrations. In specific embodiments, the methods and systems of this disclosure employ some or all of the following: (1) one or more demographic measurements; (2) one or more patient assessments; (3) one or more vital signs; (4) one or more hematological laboratory values; (5) one or more chemical laboratory values, and / or (6) one or more sepsis-related biomarker concentrations. Examples of demographic measurements include at least age. Examples of patient assessments include at least the Glasgow Coma Scale (GCS). Examples of vital signs include systolic blood pressure, diastolic blood pressure, body temperature, respiratory rate, heart rate, oxygen saturation (SpO2), and / or inhaled oxygen concentration (FiO2). Examples of hematological laboratory values ​​include white blood cell count, platelet count, lymphocyte count, and / or neutrophil count. Examples of chemical laboratory values ​​include creatinine, blood urea nitrogen, potassium, chloride, total carbon dioxide (TCO2), sodium, albumin, and / or bilirubin. Examples of sepsis-related biomarkers include C-reactive protein, procalcitonin, and / or lactate. Other examples of input characteristics are provided in Table 8-13.

[0050] In certain embodiments, combinations of patient characteristics are measured, determined, analyzed, obtained, and / or considered to assess the presence or risk of having sepsis (or a stage of sepsis). In some embodiments, combinations of patient characteristics are measured, determined, analyzed, obtained, and / or considered to determine the risk of developing sepsis within the next 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48 hours or longer. In some cases, a combination of patient characteristics is measured, determined, analyzed, obtained, and / or considered to determine the risk of sepsis occurring within the next 1-48, 1-36, 1-24, 1-12, 1-6, 1-2, 2-48, 2-36, 2-24, 2-12, 2-6, 6-48, 6-36, 6-24, 6-12, 12-48, 12-36, 12-24, 24-48, 24-36, or 36-48 hours.

[0051] In some aspects, patient characteristics as input features may or may not be measured, determined, analyzed, obtained, and / or considered in a specific order (e.g., a specific pre-set order). In certain cases, the measurement, determination, analysis, obtaining, and / or consideration of a patient's characteristics may differ from the measurement, determination, analysis, obtaining, and / or consideration of another patient's characteristics. In some embodiments, one or more features are measured, determined, analyzed, obtained, and / or considered prior to one or more other features. In some embodiments, the nature of one or more features triggers consideration or a need for consideration of one or more other features, and this process can occur in real time. In one embodiment, determining whether an individual has sepsis, is at a specific stage of sepsis, or is at risk of developing sepsis (including within the aforementioned specific timeframe) may be performed only or not only after measuring, determining, analyzing, obtaining, and / or considering at least, exactly, or no more than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, or 58 or more patient characteristics.

[0052] In a particular embodiment, one or more of the following patient characteristics are measured, determined, analyzed, obtained, and / or considered to determine whether a patient has sepsis, is at risk of developing sepsis, or is in a particular stage of sepsis: age, albumin, total bilirubin, blood urea nitrogen, C-reactive protein, chloride, creatinine, diastolic blood pressure, heart rate, lactate, lymphocytes, neutrophils, platelets, potassium, procalcitonin, respiratory rate, sodium, SpO2, systolic blood pressure, body temperature, total carbon dioxide, and white blood cell count. In specific embodiments, one or more of the following patient characteristics are measured, determined, analyzed, obtained, and / or considered to determine whether a patient has sepsis, is at risk of developing sepsis, or is in a specific stage of sepsis: age, albumin, angiopoietin-1, angiopoietin-2, total bilirubin, blood urea nitrogen, C-reactive protein, CCL2 / MCP-1, CCL20 / MIP-3α, CCL4 / MIP-1β, chloride, creatinine, CX3CL1 / chemokine, CXCL10 / IP-10, diastolic blood pressure, E-selectin, factor III / tissue factor, Flt-3 ligand, G-CSF, GM-CSF, and granzyme. B. Heart rate, IFN-α, IFN-γ, IL-10, IL-15, IL-1β / IL-1F2, IL-1ra / IL-1F3, IL-2, IL-4, IL-6, IL-7, IL-B / CXCL8, lactate, leptin, lipotransferrin-2 / NGAL, lymphocytes, neutrophils, PD-L1 / B7-H1, pentameric protein-3, platelets, potassium, procalcitonin, respiratory rate, sodium, SpO2, systolic blood pressure, body temperature, TGF-α, thrombomodulin, TNF-α, total carbon dioxide, TRAIL, TREM-1, VCAM-1, VEGF, white blood cell count. Other sections of this article (e.g., any of Tables 8-13) provide a variety of available patient characteristics.

[0053] In certain embodiments, patient features are input feature values ​​for machine learning algorithms used for sepsis risk diagnosis, sepsis stage determination, or sepsis risk prediction. In at least some embodiments, the provided system initiates monitoring of input feature values ​​from one or more measurement devices or user input devices upon receiving a request that triggers the machine learning model to initiate a process of generating a determination or prediction (e.g., from a clinician's input device). In other words, in these embodiments, the provided system may not continuously monitor input feature values ​​to the point of always receiving them, but may monitor input feature values ​​only after receiving a request until the machine learning model generates a prediction or determines that the machine learning model cannot generate a prediction based on the available input feature values. In other embodiments, the provided system may continuously monitor input feature values. Input feature values ​​collected by measurement devices or user input devices are received at the provided system (e.g., a computing device) (e.g., nearly instantly) when the input feature values ​​are collected (e.g., measured), such that the provided system receives input feature values ​​in real time. Upon collection of input feature values, triggering logic determines whether and when all input features of the machine learning model have available values ​​in order to generate a prediction, which may include interpolating values ​​to the input features.

[0054] III. Algorithm Description and Exemplary Workflow Overview In a specific embodiment, the sepsis immune scoring algorithm is a process of generating a sepsis immune score and its auxiliary components using a real-time measurement dataset. In this specific embodiment, the algorithm's inputs (including, for example, the algorithm's core inputs) are real-time measurement data stored in a database. The algorithm's core outputs may include the following: 1. Sepsis Immune Score, Sepsis Risk Score 2. Sepsis Immune Scoring Risk Stratification Categories 3. Input Features a. value b. Interpolation (true / false) c. SHAP value In a particular embodiment, the algorithm includes multiple functional components and artifacts that work together to generate an output (which can be viewed as a sepsis immune score), and also includes its auxiliary components. These functional components and artifacts reside within the backend and R engine modules of SaMD, detailed in U.S. Provisional Patent Application No. 63 / 459,521, filed April 14, 2023, the entire contents of which are incorporated herein by reference. A visual description of an example of the sepsis immune score algorithm process can be found in... Figure 1 Found it.

[0055] right Figure 1 A brief description of some functional components or artifacts of at least one example of the algorithm shown can be found in Table 1 and is described after the table.

[0056] Table 1: Overview Example of Sepsis Immune Scoring Algorithm Components – Overview of Sepsis Immune Scoring Algorithm Components i. Streaming data and input characteristics In certain embodiments, the sepsis immune scoring algorithm receives and uses a variety of different streaming data features, including one or more of the following: demographic measurements, patient assessments, vital signs, hematological laboratory values, chemical laboratory values, and concentrations of sepsis-related biomarkers. In some examples, these input features may be derived from clinical observations and / or FDA-approved devices.

[0057] In a specific embodiment, the sepsis immune score uses LOINC codes to identify streaming data features. For each feature, a base LOINC code can be provided in the software. Due to the dynamic nature of the LOINC database and differences in how input features are implemented, in some embodiments, the association between input features and LOINC codes can be updated post-release during the installation and deployment process.

[0058] In specific embodiments, the algorithm's triggering logic can be used to transform streaming data features into input features. Each input feature may have an associated input feature selection method that specifies how to select input features from the streaming measurement dataset. In some embodiments, the input feature selection method is planned to be fixed after the publication of the Sepsis Immunological Score (SaMD). A detailed description of this process is provided in U.S. Provisional Patent Application No. 63 / 459,521, filed April 14, 2023, the entire contents of which are incorporated herein by reference. Descriptions of examples of input features are listed in Tables 2 and 8-13.

[0059] Table 2: Examples of Input Features – A list of examples of input features and measurement types used in the sepsis immune scoring algorithm.

[0060] In a specific embodiment, the output is used to categorize an individual into one of four risk stratification categories: low risk, intermediate risk, high risk, or very high risk. The risk stratification category may be correlated with the incidence of the following adverse clinical events relative to the time of prescription for sepsis immune score: 1. In-hospital mortality rate 2. Readmission within 30 days 3. Transfer to ICU within 24 hours for care. 4. Start mechanical ventilation within 24 hours. 5. Use vasopressors (e.g., norepinephrine, vasopressin, epinephrine, dopamine, angiotensin II, terlipressin, selenopressin) within 24 hours. The stratification bands can be defined by three thresholds determined during the development of the sepsis immune scoring algorithm. The calculation used to identify risk stratification categories is as follows: In the development of the sepsis immune scoring algorithm, receiver operating characteristic (ROC) curves were generated using out-of-bag uncalibrated predictions and training labels. These ROC curves were used to determine three thresholds based on the following criteria: 1. Minimum threshold – Determine the threshold that achieves the highest sensitivity while maintaining a 50% false positive rate.

[0061] 2. Optimal threshold – The threshold is determined to be the “top left” point of the ROC curve, that is, the point on the ROC curve with the smallest Euclidean distance to (0,1).

[0062] 3. 95th percentile threshold – The threshold corresponding to the 95th percentile of the sepsis risk score in the training dataset.

[0063] In a specific embodiment, possible output values ​​are provided in Table 3.

[0064] Table 3: Examples of Output Values Table 4 provides examples of diagnostic interpretations and recommended actions for each potential risk stratification category as output.

[0065] Table 4: Examples of Risk Stratification Categories and Related Scores When the sepsis risk score is exactly equal to the threshold (e.g., 12.2%), it will be classified into a higher risk stratification category.

[0066] * * * * * * * * * * Various embodiments assess the presence or risk of sepsis in an individual based on one or more factors included in any of Tables 8-13 (which, given their diverse nature, may be referred to as input features, such as demographic information, patient assessment, vital signs, hematological laboratory measurements, chemical panel measurements, sepsis-related biomarkers, etc.), one or more factors included in any of Tables 8-13 (which, given their proteinic nature, may be referred to as biomarkers), or both. In some embodiments, an output is generated using one or more factors included in any of Tables 8-13 that determines the presence or risk of sepsis.

[0067] III.A. Examples of Input Features In a particular embodiment, one or more of the following input features are used to assess whether an individual has sepsis or is at risk of developing sepsis, for example, within 24 hours after measurement.

[0068] i. Demographic measurements Age can be calculated as [current date - date of birth]. Age has historically been considered a risk factor because the incidence of sepsis increases with age. In specific embodiments, this may be associated with an increase in chronic diseases that are crucial to the progression or occurrence of sepsis.

[0069] ii. Patient assessment The Glasgow Coma Scale (GCS) is a cognitive impairment assessment performed by certified professionals. GCS scores range from 3 to 15 and are assessed by healthcare professionals. The GCS is designed to assess the central nervous system component of the Sequential Organ Failure Assessment (SOFA) score (Jones et al., Critical Care Medicine, May 2009; 37(5): 1649–1654, the entire contents of which are incorporated herein by reference). The SOFA score is used to monitor a patient's condition to determine the extent or rate of organ failure. The score is based on six distinct scores corresponding to the respiratory, cardiovascular, hepatic, coagulation, renal, and nervous systems. In a specific embodiment, sepsis is defined as organ dysfunction (acute change in SOFA score ≥2), and septic shock is defined as organ dysfunction (acute change in SOFA score ≥2) and refractory hypotension.

[0070] iii. Vital signs 1. Systolic blood pressure Systolic blood pressure is the pressure exerted by the heart as it beats and pumps blood into the arteries. It can be measured using instruments such as a blood pressure monitor. Systolic blood pressure is used to calculate mean arterial pressure. Mean arterial pressure is designed to assess the cardiovascular component of the SOFA score.

[0071] 2. Diastolic blood pressure Diastolic blood pressure is the pressure in the arteries between two heartbeats at rest. It can be measured using instruments such as a blood pressure monitor. Diastolic blood pressure is used to calculate mean arterial pressure. Mean arterial pressure is designed to assess the cardiovascular component of the SOFA score.

[0072] 3. Body temperature Body temperature refers to the specific temperature inside the human body. Body temperature can be measured in several ways; a common method is using a thermometer. Body temperature can be used to determine if a patient has a fever. Fever patterns can be used to identify possible causes of infection, as recurrent fevers (fluctuations exceeding 1.1°C and a persistent fever that does not return to normal) are often caused by infection.

[0073] 4. Respiratory rate Respiratory rate is defined as the number of breaths per minute. It can be measured manually or using instruments such as a pulse oximeter. An abnormal respiratory rate may indicate an unstable supply of oxygen to the body. Tachypnea (increased respiratory rate) is common in sepsis patients and is associated with an increased 28-day mortality rate in sepsis patients. Respiratory rate is also associated with SIRS, which was part of the earlier sepsis-2 definition.

[0074] 5. Heart rate Heart rate refers to the number of heartbeats within a specific time frame. Heart rate can be measured manually or using instruments such as a heart rate monitor. Irregular heart rate may be associated with poor blood circulation and low or high blood pressure. Heart rate is also related to SIRS, which was part of the earlier definition of sepsis-2.

[0075] 6. Blood oxygen saturation (SpO2) SpO2 is a measurement of the amount of oxygen-carrying hemoglobin in the blood relative to oxygen-free hemoglobin. SpO2 can be calculated and measured by various methods, one common method being the use of a pulse oximeter. SpO2 is a non-invasive method used to estimate arterial oxygen saturation (SaO2), which is used to calculate PaO2 / FiO2 in the SOFA score for respiratory systems.

[0076] 7. Inhaled oxygen concentration (FiO2) FiO2 refers to the concentration of oxygen in an inhaled gas mixture. The oxygen concentration in an indoor air gas mixture is 21%. An elevated FiO2 may indicate that the patient requires a higher percentage of oxygen. FiO2 is used in calculating the PaO2 / FiO2 ratio in the SOFA (Solar Oxygen Deficiency Assessment) score for respiratory system function.

[0077] iv. Hematology laboratory values 1. White blood cell count White blood cell count is a measurement of the number of white blood cells in a patient's blood. It can be measured through a hematological laboratory test. An abnormal white blood cell count may indicate that a patient is fighting an infection.

[0078] 2. Platelet count Platelet count is a measurement of the number of platelets in a patient's blood. Platelet count can be measured through a hematological laboratory test. An abnormal platelet count may indicate a coagulation disorder in an individual. Platelets are involved in hemostasis and immune responses and play a crucial role in sepsis. Platelet count is used to assess the coagulation system component of the SOFA score.

[0079] 3. Lymphocyte count A lymphocyte count is a measurement of the number of lymphocytes in a patient's blood. Lymphocyte counts can be measured through a hematological laboratory test. Lymphocytes are a type of white blood cell and are part of the immune system. An abnormal lymphocyte count may indicate the presence of an infection. The combination of lymphocytes and neutrophils can be used to calculate the neutrophil-to-lymphocyte ratio (NLR). NLR is associated with mortality in sepsis patients because non-survivors have significantly higher NLRs.

[0080] 4. Neutrophil count Neutrophil count is a measurement of the number of neutrophils in a patient's blood. Neutrophil count can be measured through a hematological laboratory test. Neutrophils are granulocytes, a type of white blood cell, and are part of the immune system. Abnormal lymphocyte counts may indicate the presence of an infection. The combination of lymphocytes and neutrophils can be used to calculate the neutrophil-to-lymphocyte ratio (NLR). NLR is associated with mortality in sepsis patients because non-survivors have significantly higher NLRs.

[0081] v. Chemical laboratory value 1. Creatinine Creatinine is defined as the concentration of creatinine in the blood. Blood creatinine measurements can be performed using a chemistry lab panel. Elevated blood creatinine indicates that the renal system may be unable to filter creatinine into the urine for clearance. Creatinine is used to assess the renal system portion of the SOFA score.

[0082] 2. Blood urea nitrogen Blood urea nitrogen (BUN) refers to the amount of residual urea nitrogen in the blood. Blood BUN can be measured using a chemistry lab panel. Urea nitrogen is a metabolic waste product that is transferred from the blood to the urine by the kidneys. High BUN may indicate that the kidney system may not be functioning effectively. Elevated BUN is associated with the occurrence or presence of sepsis.

[0083] 3. Potassium Potassium refers to the concentration of potassium ions in the blood. Blood potassium levels can be measured through chemical laboratory tests. Potassium plays a crucial role in maintaining fluid levels within the body's cells. Potassium imbalances can lead to fluid level imbalances. Fluid disturbances are associated with many patients' admission to the ICU, and fluid resuscitation is highly effective in treating patients with severe sepsis or septic shock.

[0084] 4. Chloride Chloride refers to the concentration of chloride ions in the blood. Blood chloride levels can be measured through chemical laboratory tests. Chloride plays a crucial role in maintaining fluid levels within the body's cells. Chloride imbalances lead to fluid level imbalances. Fluid level disturbances are associated with many patients' admission to the ICU, and fluid resuscitation is highly effective in treating patients with severe sepsis or septic shock.

[0085] 5. Total carbon dioxide (TCO2) Total carbon dioxide (TCO2) is a measurement of carbon dioxide found in the blood. TCO2 can be measured in the blood through chemical laboratory testing. TCO2 can be used to identify metabolic acidosis, which is related to the respiratory system. TCO2 imbalances may be associated with respiratory dysfunction. Elevated TCO2 measurements are associated with increased mortality in patients with sepsis.

[0086] 6. Sodium Sodium refers to the concentration of chloride ions present in the blood. Blood sodium levels can be measured through chemical laboratory tests. Sodium plays a crucial role in maintaining fluid levels in the body's cells. Sodium imbalance leads to fluid level imbalance. Fluid level disturbances are associated with many patients' admission to the ICU, and fluid resuscitation is an effective treatment for patients with severe sepsis or septic shock.

[0087] 7. Albumin Albumin refers to the concentration of albumin present in the blood. Blood albumin levels can be measured through a chemical laboratory test. Albumin is a protein synthesized by the liver, which minimizes leakage from capillaries into body tissues. Abnormal albumin levels may indicate problems with the liver or kidneys. Serum albumin has been shown to be a predictor of mortality in patients with sepsis.

[0088] 8. Bilirubin Bilirubin refers to the concentration of bilirubin present in the blood. Blood bilirubin levels can be measured through chemical laboratory tests. Bilirubin is a protein produced during the breakdown of red blood cells. In healthy individuals, bilirubin is a metabolic waste product of the liver. Bilirubin is used in the SOFA score calculation to assess the liver portion of the test.

[0089] vi. Concentration of sepsis-related biomarkers 1. C-reactive protein C-reactive protein (CRP) refers to the concentration of CRP in the blood. Blood CRP levels can be measured through a chemical laboratory test. CRP is a protein produced by the liver, and its levels rise due to inflammation. Studies have shown that using CRP in conjunction with body temperature can improve the detection rate of infections.

[0090] 2. Procalcitonin Procalcitonin (PCT) refers to the concentration of PCT present in the blood. PCT can be measured by laboratory tests. During inflammation, PCT is released from tissues into the bloodstream. Bacterial infection has been shown to have a significant impact on PCT synthesis because bacterial endotoxins and cytokines promote the secretion of PCT by host cells.

[0091] 3. Lactic acid Lactate refers to the concentration of lactate present in the blood. Lactate can be measured through laboratory tests. Lactate (also known as lactic acid) is produced by tissues under hypoxic conditions. Accumulation of lactate in the blood may indicate organ failure or shock. In the sepsis-3 definition, lactate has become part of the labeling for septic shock.

[0092] In some embodiments, individuals at risk of or suspected of having sepsis have had one or more samples analyzed, measured, considered, and / or determined. Examples of samples include blood, plasma, serum, urine, tissue, skin, saliva, sputum, and / or mucus. The level of any one or more patient characteristics in the sample may be used as input features in the model described herein. Alternatively, or additionally, individuals may undergo blood pressure monitoring, imaging studies (X-ray, ultrasound scan, or computed tomography (CT) scan), etc.

[0093] In some embodiments, multiple input features are selected from 1, 2, 3, or more categories. These features may or may not be used in any model described herein. In specific embodiments, the multiple input features include at least one input feature from a first group, at least one input feature from a second group, and at least one input feature from a third group. In some embodiments, the first group includes input features from the group consisting of: age, sex at birth, race, ethnicity, patient history, current patient complaint or symptoms, neurological assessment, clinical decision support alerts, clinician or other chart records, diagnostic codes, surgeries performed on the patient, current patient medications, interventions, patient care environment, medical imaging data or assessments, electrocardiogram, endoscopy, systolic blood pressure, diastolic blood pressure, body temperature, respiratory rate, heart rate, blood oxygen saturation, inhaled oxygen concentration, and combinations thereof. In some embodiments, the second group includes input features from the group consisting of: white blood cell count, lymphocyte count, neutrophil count, platelet count, creatinine, blood urea nitrogen, potassium, chloride, total carbon dioxide, sodium, albumin, bilirubin, lactate, glucose, calcium, hematocrit, hemoglobin concentration, laboratory-derived biomarkers, clinical severity indicators and composite indicators, biopsy data, microbial test data, and combinations thereof. In some embodiments, the third group includes input features from the group consisting of: procalcitonin, C-reactive protein, protein biomarkers, genomic biomarkers, gene expression or transcriptome biomarkers, composite biomarker scores, and combinations thereof.

[0094] In a specific embodiment, the following categories exist, and one or more input features from any one or more categories can be used to determine whether an individual has sepsis or is at risk of developing sepsis within a predetermined time: Category 1 • Vital signs (e.g., body temperature, heart rate, blood pressure, respiratory rate, blood oxygen saturation) • Demographic information (e.g., age, sex, race, ethnicity) • Past medical history • Current chief complaint or symptoms • Neurological assessment (Glasgow Coma Scale, Complete Unresponsiveness Assessment (FOUR)) • Clinical decision support alerts • Clinician or other chart records • Diagnostic codes (e.g., ICD-10) • Surgery (including CPT codes) • Current medication (including NDC code) • Intervention measures • Patient care environment (ICU, ED, wards, outpatient department, etc.) • Imaging examinations (e.g., X-ray, CT scan, MRI, fMRI, ultrasound) • Electrocardiogram (e.g., EEG, EKG) • Endoscopy Category 2 • Laboratory tests • Lab-derived biomarkers (e.g., monocyte distribution width, Cytovale IntelliSep) • Clinical severity indicators and comprehensive indicators (e.g., Sequential Organ Failure Assessment (SOFA) score, SOFA component scores, Charlson Comorbidity Index, Acute Physiology and Chronic Health Assessment (APACHE) I and II). • Biopsy • Microbial testing (culture, PCR testing, antigen testing) Category 3 • Protein biomarkers (including angiopoietin-1, angiopoietin-2, C-reactive protein, cystatin C, D-dimer, E-selectin, chemokines, FLT3 ligand, GCSF, GDF15, GMCSF, granzyme B, IFN-α, IFN-γ, IL1-β, IL1-RA, IL-2, IL-4, IL-6, IL-7, IL-8, IL-10, IL-15, IP-10, lactate dehydrogenase (LDH), lipopolysaccharide-binding protein (LBP), leptin, MCP1, MIP1-α, MIP1-β, MIP3-α, NGAL, islet calculiin, PDL-1, pentaprotein-3, procalcitonin, protein C, Sl00b, TGF-α, thrombomodulin, tissue factor, TNF-α, TRAIL, TREM-1, troponin, VCAM-1, VEG-F) • Genomic biomarkers (e.g., one or more SNPs from CD14, TLR1, TLR2, TLR4, TLR6, TNF-α, IL-6, factor V, factor XII, FER, MAN2A1) • Gene expression or transcriptome biomarkers (including CEACAM4, LAMP1, PLAC8, PLA2G7, CEACAM1, ZDHHC19, C9orf95, GNA15, BATF, C3AR1, KIAA1370, TGFBI, MTCH1, RPGRIP1, HLA-DPB1, nuclear factor-κB (NF-κB), heme oxygenase-1 (HO-1), FCGR2C, CRP, IL-6, IL-10, IL-12, TNF, LY6E, LAMP3, ISG15, USP18, RSAD2, IFI44L, NCOA7, CMPK2, IFI44, BATF2, GBP4, I FIH1, GMPR, CXCL10, IFI27, ISG15, CCL2, OAS3, P2RY14, GCH1, CD274, ID01, ARL4A, BPGM, TNFAIP3, TNFAIP6, IL1R1, PROK2, RGS1, CXCR6, RTP4, TLR3, FZ D5, KCNJ2, IRAK2, RGS16, CXCL8, OLR1, CCRL2, SPP1, ST3GAL4, CST7, IL1R2, IL4R, GALM, CTLA4, SOCS1, BCL2L1, TGM2, SLC39A8, CA2, PNP, GATA1, SLC1A5, RGS16, CD44, PDGFC, CSF2RA, CXCL1, CSF3R, SOCS3, IL18R1, PIM1, LEPR, IL1B, CSF2RB, ITGB3, STAT1, FAS, EFNA1, IER3, EETS2, SLC2A3, BCL2A1, PFKFB3 , CEBPB, BIRC2, ATF3, TUBB2A, G0S2, MERTK, BIK, QSOX1, PYGL, TGFA, P4HA2, LDHA, IRS2, TSPO, HGF, GADD45A, TGFBR3, CD38, PRF1, FASLG, TIMP3, ANKH, LG ALS3, PPP2R5B, GPX1, BNIP3L, BCL2L1, KLF7, FOSL2, ADM, MXI1, SELENBP1, PGF, FOXO3, NEDD4L, SLC2A1, SIAH2, MMP9, CXCR3, CD8A, IFNG, CD8B, JAK2, CC L4, ICAM1, WARSI, KRT1, GPR65, DYRK3, ACHE, ARG1, LCN2, LTF, OLFM4, HLA-DMB, YKT6, PDE4B, TWISTNB, BTN2A2, ZBTB33, PSMB9, CAMK4, TMEM19, SLC12A7,TP53BP1, PLEKHO1, SLC25A22, FRS2, CD24, STX1A, KCNMB4, CRISP2, HTRA1, PPL, RHBDF2, ZCCHC4, DDX6, SENP5, RAPGEF1, DTX2, RELB, DYRK2, CCNB1IP1, TDRD9, ZAP70, ARL14EP, MDC1, ADGRE3), • Composite biomarker scores (e.g., SeptiCyte LAB and Septicyte RAPID, Inflammatix TruVerity) III.B. Algorithm Output In a specific embodiment, the core output of the sepsis immune scoring algorithm is as follows: 1. Sepsis Immune Score, Sepsis Risk Score 2. Sepsis Immune Scoring Risk Stratification Categories 3. Input Features a. value b. Interpolation (true / false) c. SHAP value i. Sepsis Immune Score In a particular embodiment, the sepsis immune score and sepsis risk score are continuous values ​​between 0 and 100 and are the output of the calibrated prediction function described elsewhere in this document. In a particular embodiment, the sepsis immune score is intended to be displayed to SaMD users. In some embodiments, the sepsis immune score can be interpreted as the probability that a patient has or develops sepsis (as defined by Sepsis-3) within 24 hours from the date of the sepsis immune score prescription. In a particular embodiment, the sepsis immune score can be differentiated from an uncalibrated sepsis risk score by applying calibration that improves the interpretability of the uncalibrated sepsis risk score.

[0095] ii. Risk Category In a specific embodiment, the sepsis immune score is used to categorize patients into one of four risk stratification categories: low risk, intermediate risk, high risk, or very high risk. The risk stratification category is associated with the incidence of the following adverse clinical events relative to the time to prescription based on the sepsis immune score: 1. In-hospital mortality rate 2. Readmission within 30 days 3. Transfer to ICU within 24 hours for care. 4. Start mechanical ventilation within 24 hours. 5. Use vasopressors within 24 hours. In some embodiments, risk stratification categories are determined by three thresholds identified during the development of the sepsis immune scoring algorithm. The calculations used to identify risk stratification categories are described elsewhere in this document.

[0096] iii. Input Features In a particular embodiment, the sepsis immune score retains the values ​​of the input features used to calculate the sepsis immune score, a TRUE or FALSE value indicating whether the value is an imputed value, and the associated SHAP value. The SHAP value is a numerical value representing the local feature importance of the sepsis immune score to a specific set of input features. In some respects, using the SHAP value helps in interpreting and trusting the sepsis immune score. In a particular embodiment, the sum of the SHAP values ​​of individual sepsis immune scores equals the uncalibrated sepsis immune score prediction—the average of a predefined dataset (the pre-calibrated sepsis immune score prediction). Thus, the cumulative effect of each input feature on the pre-calibrated sepsis immune score prediction can be observed.

[0097] The SHAP value is not a global algorithmic property, but rather a property of any newly generated score, and in some embodiments, the commonly used concept of feature importance is localized to a single score.

[0098] III.C. Functional Components iv. Triggering Logic enter Streaming data characteristics Output Can the sepsis immune score generate results? Input features with missing values Triggering logic can identify whether an active sepsis immune score command is ready to generate a sepsis immune score. In some embodiments, once the triggering logic indicates that a sepsis immune score can be generated, it passes input features to the interpolation function component. More information regarding the specification and description of the triggering logic can be found in U.S. Provisional Patent Application No. 63 / 459,521, filed April 14, 2023, the entire contents of which are incorporated herein by reference.

[0099] v. Interpolation input features enter Input features with missing values Output Input features without missing values workpiece interpolation template The input feature interpolation function receives input features identified from the trigger logic module to interpolate any missing values. In a specific embodiment, the input feature interpolation may perform the following steps in sequence: 1. Interpolate the missing FiO2 measurements to 21. 2. Interpolate the missing GCS measurements to 15. 3. Perform bag interpolation using the interpolation template.

[0100] For more information on interpolation methods, please refer to other sections of this article.

[0101] vi. Generate uncalibrated predictions enter Input features; missing values ​​have been imputed.

[0102] Output Uncalibrated sepsis immune score. Uncalibrated original model predicted score.

[0103] workpiece Machine learning models In a particular embodiment, the pre-calibration prediction generation function uses the output from the interpolation input feature function and a trained machine learning model to generate an original pre-calibration sepsis immune score, which is a continuous value between [0,1]. The machine learning model used may be a random forest model, implemented using the Ranger R package. More information about this machine learning model can be found in other sections of this document.

[0104] vii. Calibrate prediction enter Uncalibrated sepsis risk score Output Sepsis Risk Score workpiece Calibration Model The calibration prediction function uses an uncalibrated sepsis immune score and a calibration model to generate a sepsis immune score. More information about the calibration model can be found in other sections of this document.

[0105] viii. Generate SHAP value enter Input data (interpolation) Output SHAP value workpiece Machine learning models Training data objects In some embodiments, the SHAP value generation function uses interpolated input features, a trained machine learning model, and training objects to generate SHAP value contributions of the input features. Specifically, the computation of SHAP values ​​employs game theory methods to identify the role of features in predicting observations (Scott, Lundberg M, and Lee Su-In, “A Unified Approach to Explaining Model Predictions,” 31st Conference on Neural Information Processing Systems, 2017). In specific embodiments, SHAP values ​​can use training data objects to identify estimated contributions of features to the training dataset. Due to the computational complexity of SHAP value computation, in at least some cases, the SHAP value generation function can be implemented using SHAP estimation via the fastshap R package. In some embodiments, the fastshap R package uses a 10-round Monte Carlo simulation to estimate the feature contributions of training data objects with a fixed seed. In some embodiments, this estimation can be applied to new observations to generate an overall contribution to the prediction.

[0106] ix. Generate risk stratification categories enter Sepsis Risk Score Output Risk stratification categories workpiece Threshold list In some embodiments, the risk stratification category generation function uses a pre-established list of thresholds to categorize sepsis immune score and sepsis risk score values ​​into one of four categories. The risk stratification category generation function categorizes patients into risk categories according to the following logic: Table 5: Generation of Risk Stratification Categories III.D. Workpiece The sepsis immune scoring algorithm includes five artifacts in some embodiments: 1. Interpolation template 2. Pre-trained machine learning algorithms 3. Calibration Model 4. Training data objects 5. Threshold List All artifacts can be created during the development of the sepsis immune scoring algorithm. In a particular embodiment, they are intended to be fixed after the release of the sepsis immune score SaMD.

[0107] i. Interpolation template In a specific embodiment, the interpolation template is a pre-trained template used to interpolate values ​​for missing input features. The interpolation template can be stored as an R data structure file. In one embodiment, the purpose of this artifact is to ensure that all input features have usable measurements before being used to generate an uncalibrated sepsis risk score.

[0108] ii. Pre-trained machine learning models In some embodiments, the pre-trained machine learning model is a supervised machine learning model constructed using a training dataset containing input features and training labels, along with hyperparameters. In certain aspects, the machine learning model is stored as an R data structure file. The purpose of the machine learning model is to learn the relationship between input features and the risk of developing or being diagnosed with sepsis within 24 hours of the sepsis immune score prescription. In at least some cases, this relationship can be applied to future unobserved patients with the corresponding input features.

[0109] iii. Calibration Model In a particular embodiment, the calibration model is a logical model that takes an uncalibrated sepsis risk score as input to generate a sepsis immune score / sepsis risk score. The calibration model can be saved as an R data structure file. In one embodiment, the purpose of the calibration model is to establish a relationship between an uncalibrated sepsis risk score and the classification of having or developing sepsis within 24 hours of a sepsis immune score prescription. In some embodiments, this relationship can be used to convert future uncalibrated sepsis risk scores into sepsis risk scores.

[0110] iv. Training data objects In a particular embodiment, the training data object is a data frame containing input features of a training dataset. The training data object can be saved as an R data structure file. In some embodiments, the purpose of the training data object is to provide data to learn the relationship between the SHAP values ​​of the input features and their corresponding uncalibrated sepsis risk score values. In some embodiments, this relationship can be applied to future sepsis immunization score prescriptions to generate SHAP values ​​corresponding to the sepsis risk scores.

[0111] v. Threshold list In some embodiments, the threshold list is a vector containing three thresholds between 0 and 1, which are associated with the boundaries of the sepsis immune score risk categories. The threshold list can be stored in a JSON file. In a specific embodiment, these three values ​​will be ordered in ascending order and referred to as the "lowest," "best," and "95th percentile" thresholds, respectively.

[0112] III.E. Datasets Used for Development The algorithm was developed using retrospective data of de-identified patient data and samples derived from the multicenter NOSIS dataset and biobanks.

[0113] In a specific embodiment, the NOSIS dataset and the biobank are unified datasets containing prospectively collected clinical data (electronic health record data), time-series biosamples, and biomarker measurements obtained from these samples. This dataset was constructed through prospective clinical studies conducted by the applicant in a consortium of multiple clinical centers.

[0114] III.F. Algorithm Artifact Design The following sections define the methods for designing each function and artifact used in the sepsis immune scoring algorithm.

[0115] i. Interpolation template In some embodiments, the imputation template is generated using bag imputation on the training dataset of the sepsis immune scoring algorithm described elsewhere in this document. Bag imputation is a statistical method that, in specific embodiments, constructs a random forest model for each input feature in the sepsis immune scoring algorithm. Each random forest model can generate imputed values ​​using the remaining observed input features. In some embodiments, the imputation template can be evaluated by performing imputation on unseen data and calculating R² of the imputed measurement for each input feature with respect to known measurements.

[0116] This demonstrates how bag interpolation is implemented using the imputeMissings R package.

[0117] In some embodiments, the data used to develop the interpolation template is the same as the data used for the machine learning model.

[0118] ii. Pre-trained machine learning models Pre-trained machine learning models can be developed by training a supervised machine learning model using a training dataset containing input features and training labels, combined with hyperparameters associated with the model. In some embodiments, the pre-trained machine learning model is evaluated by examining specific performance metrics related to diagnostic and prognostic performance.

[0119] iii. Machine Learning Models In a specific embodiment, the machine learning model used is a random forest model. Random forests are popular ensemble models that combine many simple tree models to generate predictions (Breiman, Leo, “Random Forests,” Machine Learning 45, 2001, pp. 5–32). Random forests perform a bagging method, which is a method of sampling a dataset with replacement. A single simple model can be trained based on this sampled dataset. In a specific embodiment, this sampling with replacement and subsequent training are performed n times to generate an ensemble model, i.e., a forest, composed of simple models. In some embodiments, the random forest used to develop the sepsis immune scoring algorithm used 1000 regression tree models as base models to generate a probabilistic forest (Malley, James D., Jochen Kruppa, Abhijit Dasgupta, Karen G. Malley, and Andreas Ziegler, “Probabilistic Machines,” Medical Information Methods 51, Vol. 01, 2012, pp. 74–81). In a specific implementation, the random forest model is trained by maximizing the area under the receiver operating feature curve (AUROC) between the predicted values ​​and the training labels.

[0120] The advantage of random forest models lies in their use of out-of-bag (OOB) error. Because a bagging method is used to generate each decision tree, there is a certain probability that observations were not included in the training of the decision tree. In some embodiments, this set of unincluded observations can be defined as OOB error. In specific embodiments, by generating predictions based on OOB, the error of all observations can be identified without relying on traditional training / test splits.

[0121] This implementation of a random forest is carried out using the ranger R package.

[0122] iv. Training dataset The training dataset is a subset of data from the NOSIS dataset and the biobank, and was extracted using the following criteria: This was a multicenter prospective observational study of patients suspected of having sepsis. Patients were screened according to inclusion criteria by querying the hospital's electronic medical record system. Patients with relevant medical orders indicating suspected bacteremia (blood culture) were eligible for inclusion. A list of eligible patients was created, and researchers or hospital staff assessed the presence of residual samples from lithium heparin gel separation (light green cap) tubes in the clinical laboratory. Patients with available and collected samples were included as part of the NOSIS dataset.

[0123] Adult patients (18 years and older) who presented to the emergency department or inpatient ward with suspected sepsis were included. Suspected sepsis was inferred from blood culture orders. Patient data were extracted from their hospital electronic health records (EHRs), including demographic data, comorbidities, vital signs, laboratory test results, interventions recorded in the EHRs, and ICD-10 codes. Additionally, residual plasma samples collected during routine clinical procedures (each sample consisting of four 175 µL aliquots) were collected, frozen at -80°C, and transported to the central laboratory. Each sample included at least one plasma aliquot, prospectively obtained during hospitalization and frozen at -80°C, from a maximum of 3 days prior to enrollment.

[0124] For a subset of patients in the NOSIS biobank, approximately 40 protein biomarkers, including CRP and PCT, were measured from one plasma waste sample from each patient. Based on patient enrollment date, samples closest to baseline were selected sequentially from each research center; baseline was defined as the time of the blood culture order. Patients whose biobank samples were not collected within 3 hours of their initial blood culture order were excluded. Protein concentrations were measured using the Luminex MagPix multiplex assay developed by R&D Systems. Laboratory quality control procedures strictly followed FDA guidelines for bioanalytical method validation to ensure data quality. This study recruited 2,367 patients from three hospitals, who were identified as eligible for triggering logic to serve as data for training the algorithm.

[0125] v. Input Feature Selection The selected measurements used for the input features conform to the logic provided in U.S. Provisional Patent Application No. 63 / 459,521, filed April 14, 2023, the entire contents of which are incorporated herein by reference. Interpolation of the measurements was performed using the same criteria as described elsewhere in this document. During development, the alternative to the sepsis immune score order was replaced with the initial blood culture order during the visit.

[0126] vi. Training labels The labels used to train the sepsis immune score machine learning model are sepsis-3 labels derived within 24 hours of the initial blood culture order. In a specific embodiment, the derived sepsis-3 labels are calculated based on the presence of organ dysfunction and infection. If the patient meets the infection criteria, the timing of the organ dysfunction event corresponds to the sepsis-3 timing. Further details regarding the organ dysfunction portion and the infection criteria are listed below.

[0127] vii. Organ dysfunction Organ dysfunction is calculated using existing medication information, laboratory test results, vital signs, and patient assessment data from the NOSIS database during the patient's visit. Organ dysfunction is calculated using the following formula: SOFA – SOFA baseline >= 2 The SOFA score is calculated as the sum of the SOFA components of the CNS, cardiovascular, respiratory, renal, coagulation, and hepatic systems. The methods used to calculate the SOFA score are indicated in Table 6 below. Baseline SOFA is calculated as the SOFA components at the time of the previous visit (if available). If no previous visit record is available, the baseline is assumed to be 0.

[0128] Table 6: SOFA score components – Calculation of SOFA score components (Lambden, S., Laterre, PF, Levy, MM, & Francois, B. (2019). SOFA score – development, utility and challenges in accurate assessment in clinical trials. Intensive Care, 23(1), 374).

[0129] a Administer adrenergic drugs at least 1 hour later (dosage unit is µg / kg) . min) viii. infection The infection criteria for the sepsis immune score are determined if one of the following criteria is met: 1. Within the two-day window period before and after the initial blood culture order, there are four or more qualified antibiotic use days (QAD). 2. Positive microbiological test or culture results, excluding known contaminants. QAD guidelines follow the description of infection in the CDC Adult Sepsis Hospital Toolkit (https: / / www.cdc.gov / sepsis / pdfs / Sepsis-Surveillance-Toolkit-Mar-2018_508.pdf). Positive microbiological testing is a supplement to the CDC Adult Sepsis Hospital Toolkit. A list of known contaminants in blood cultures is shown in Table 7.

[0130] Table 7: Known Blood Culture Contaminants—This table describes known contaminants in blood cultures (Hall, Keri K. and Jason A. Lyman, “A Recent Review of Blood Culture Contamination,” Clinical Microbiology Review, Vol. 19, No. 4, 2006, pp. 788–802, https: / / doi.org / 10.1128 / cmr.00062-05). False positive rates often represent the number of contaminated organisms.

[0131]

[0132] Known contaminants in urine culture are defined as “Lactobacillus, Corynebacterium spp., Gardnerella vaginalis, α-hemolytic streptococci, and aerobic bacteria” (Franz, Martina, and Walter H. Hörl, “Common Mistakes in the Diagnosis and Management of Urinary Tract Infections I: Pathophysiology and Diagnostic Techniques”, Nephrology, Dialysis and Transplantation, Vol. 14, No. 11, 1999, pp. 2746-2753, https: / / doi.org / 10.1093 / ndt / 14.11.2746).

[0133] ix. Hyperparameters In certain embodiments, the hyperparameters associated with the random forest model include the number of trials (mtry), minimum node size, and splitting rule. Various combinations of the number of trials, minimum node size, and splitting rule are considered. In some embodiments, the combination that yields the best AUROC result in three replicates of 5-fold cross-validation is selected as the optimal hyperparameters. The R package 'caret' is used to assist in the selection of optimal hyperparameters.

[0134] x. Performance Evaluation In some embodiments, the pre-trained machine learning model is evaluated by assessing diagnostic and prognostic performance. Both performance metrics are evaluated using a calculated sepsis risk score and a risk threshold determined during the development of the sepsis immune scoring algorithm.

[0135] xi. Risk threshold setting During the development of the sepsis immune scoring algorithm, receiver operating characteristic (ROC) curves were generated using uncalibrated prediction and training labels. In some embodiments, these ROC curves were used to determine three thresholds based on the following criteria: 1. Minimum threshold – Determine the threshold that achieves the highest sensitivity while maintaining a 50% false positive rate.

[0136] 2. Optimal threshold – The threshold is determined as the “top left” point of the ROC curve, that is, the point on the ROC curve with the smallest Euclidean distance to (0,1).

[0137] 3. 95th percentile threshold – The threshold corresponding to the 95th percentile of the sepsis risk score in the training dataset.

[0138] xii. Diagnostic performance The diagnostic performance of the global machine learning model was measured by determining the global area under the OOB AUROC and the area under the OOB precision-recall curve (AURPC). Furthermore, diagnostic metrics (e.g., sensitivity, specificity, positive predictive value, negative predictive value, etc.) were evaluated at each threshold.

[0139] xiii. Prognostic performance Prognostic performance is assessed by determining the incidence of adverse events in each of the four risk categories described in other parts of this paper.

[0140] xiv. Calibration Model While classifiers generated by random forest models are accurate and achieve high AUCs, the quality of the resulting class probabilities is quite poor, as measured by the squared error of the predicted probabilities (Brier score). One approach to address this is to use a calibration model, which learns a function that maps the raw probability estimates, or scores, to more accurate probability estimates.

[0141] In a particular embodiment, the calibration model is generated by performing Pratt calibration. Pratt calibration can be created by training a logistic regression model using a pre-calibration sepsis immune score to predict sepsis-3 training labels. The calibration model can be evaluated by determining the percentage of sepsis patients with a given sepsis immune score.

[0142] IV. Exemplary Methods IV.A. Exemplary Methods Figure 2 An example method 200 for generating a sepsis risk indicator is illustrated according to various embodiments. At step 210 of method 200, the method includes the step of receiving at least one input feature corresponding to a patient. At step 220, the method further includes the step of analyzing the at least one input feature using a machine learning model to generate a predictive score based on the at least one input feature, the predictive score indicating the probability that the patient has or develops sepsis over a predetermined time span. Optionally (step 230), according to various embodiments, at least one input feature is associated with sepsis.

[0143] According to various embodiments, at least one input feature can be divided into two groups. These two groups can be concentration values, wherein the relative importance of the first group of concentration values ​​is lower than that of the second group of concentration values.

[0144] According to various embodiments, at least one input feature includes procalcitonin and can be divided into two groups. These two groups can be procalcitonin concentration values, wherein the relative importance of the first group of procalcitonin concentration values ​​is lower than that of the second group of procalcitonin concentration values.

[0145] According to various embodiments, at least one input feature can be divided into three groups. These three groups can be concentration values, wherein the relative importance of the first group of concentration values ​​is lower than that of the second group of concentration values, and the relative importance of the third group of concentration values ​​is higher than that of the first and second groups of concentration values.

[0146] According to various embodiments, at least one input feature includes procalcitonin and can be divided into three groups. These three groups can be procalcitonin concentration values, wherein the relative importance of the first group of procalcitonin concentration values ​​is lower than that of the second group of procalcitonin concentration values, and wherein the relative importance of the third group of procalcitonin concentration values ​​is higher than that of the first and second groups of procalcitonin concentration values.

[0147] Figure 9A shows a non-limiting example of two sets or box features applied to Model 1A (discussed in detail below). Figure 9B shows a non-limiting example of three sets or box features applied to Model 1A.

[0148] For example, the example method can be further modified according to various embodiments, as provided in claims 2-171. The additional features included can be implemented in any combination as needed.

[0149] V. Example The following examples are intended to illustrate preferred embodiments of the present invention. Those skilled in the art will understand that the techniques disclosed in the following examples represent techniques discovered by the inventors and effective in the practice of the invention, and therefore can be considered preferred embodiments constituting their practice. However, those skilled in the art will understand from this disclosure that many modifications can be made to the specific embodiments disclosed without departing from the spirit and scope of the invention, and the same or similar results can still be obtained.

[0150] Example 1 Example of algorithm results and feature importance ranking for exemplary models This example provides examples of algorithmic results and feature importance rankings for multiple versions of the sepsis algorithm. The difference between these versions lies in the set of input features used to predict sepsis. In a particular embodiment, six different models utilize various of the following: one or more demographic measurements, one or more patient assessments, one or more vital signs, one or more hematological laboratory values, one or more chemical laboratory values, and / or one or more concentrations of sepsis-related biomarkers.

[0151] VA Model 1A The following are exemplary input features for Model 1A, in no particular order.

[0152] Table 8: Examples of input features for Model 1A age albumin bilirubin BUN chloride CO2 C-reactive protein Creatinine diastolic blood pressure Heart rate lactic acid lymphocytes neutrophils platelets Potassium Procalcitonin Pulse oxygen saturation respiratory rate sodium systolic blood pressure body temperature White blood cell count In one embodiment of Model 1A Figure 3A (No sepsis occurred within 24 hours) Figure 3B (Sepsis occurred within 24 hours) and Figure 3C The section "(No sepsis within 24 hours)" provides examples of SHAP values ​​for three different high-risk sepsis patients. The order of importance of the input features differs for these three patients.

[0153] Figure 3A A snapshot of Patient 1's SHAP values ​​for the listed input characteristics is shown. Most of the patient's characteristics have negative SHAP values, indicating a lower sepsis immune score, particularly given their young age, and normal laboratory values ​​for BUN, platelets, and creatinine, none of which suggest dysfunction in these organs. However, elevated bilirubin indicates abnormal liver function, while elevated respiratory rate and mild fever have significant positive SHAP values.

[0154] Figure 3B A snapshot of Patient 2’s SHAP values ​​for the listed input features is shown. Most of the features of this elderly patient have positive SHAP values, with platelet count, low pulse oxygen saturation and BUN value having the greatest impact. These values ​​all indicate organ dysfunction, which meets the definition of organ dysfunction in sepsis-3 (SOFA score increase of at least 2 points from baseline).

[0155] Figure 3CA snapshot of Patient 3's SHAP values ​​for the listed input characteristics is shown. Except for age and a few other slightly positive laboratory indicators, almost all of this patient's characteristics have negative SHAP values. Organ function and infection-related characteristics are represented by the most important SHAP values, and both are negative. This patient is unlikely to develop sepsis within the next 24 hours.

[0156] Exemplary results of the diagnostic and prognostic performance of Model 1A are as follows:

[0157] VB Model 1B The following are exemplary input features of Model 1B, in no particular order. In a particular embodiment, the input features of Model 1B include IL-6, which may be missing from the input features of Model 1A.

[0158] Table 9: Examples of input features for Model 1B age albumin bilirubin BUN chloride CO2 C-reactive protein Creatinine diastolic blood pressure Heart rate IL-6 lactic acid lymphocytes neutrophils platelets Potassium Procalcitonin Pulse oxygen saturation respiratory rate sodium systolic blood pressure body temperature White blood cell count Exemplary results of the diagnostic and prognostic performance of Model 1B are as follows: VC Model 2 The following are exemplary input features for Model 2, in no particular order.

[0159] Table 10: Examples of input features for Model 2 age albumin Angiopoietin-2 bilirubin BUN chloride CO2 C-reactive protein Creatinine diastolic blood pressure Heart rate IL-6 IL-8 IP-10 lactic acid lymphocytes neutrophils Pentanoin-3 platelets Potassium Procalcitonin Pulse oxygen saturation respiratory rate sodium systolic blood pressure body temperature White blood cell count In one embodiment of Model 2, Figure 4A (Sepsis occurred within 24 hours) Figure 4B (No sepsis occurred within 24 hours) and Figure 4C The section on "(Sepsis occurring within 24 hours)" provides examples of SHAP values ​​for three different high-risk sepsis patients. The order of importance of input features varies from patient to patient.

[0160] Figure 4A This image shows a snapshot of Patient 1's SHAP values ​​for the listed input features. Most of the patient's features show positive SHAP values, with bilirubin and platelet counts being the most significant, both indicating organ dysfunction. The features of multiple infection-related biomarkers suggest that the organ dysfunction may be caused by infection, as defined in sepsis-3.

[0161] Figure 4B A snapshot of Patient 2's SHAP values ​​for the listed input features is shown. This patient's SHAP values ​​are almost entirely negative, corresponding to an extremely low sepsis immune score. Many of the most important (i.e., the highest) values ​​are protein biomarkers, including IL-8, procalcitonin, IL-6, and pentamericin-3.

[0162] Figure 4C A snapshot of Patient 3's SHAP values ​​for the listed input characteristics is shown. This patient's SHAP values ​​exhibit both significant positive and negative values, including elevated respiratory rate and elevated levels of certain sepsis-related biomarkers such as IL-6, IL-8, and CRP. However, their normal blood pressure and platelet count contributed the most overall.

[0163] Exemplary results of the diagnostic and prognostic performance of Model 2 are as follows: VD Model 3 The following are exemplary input features for Model 3, in no particular order.

[0164] Table 11: Examples of input features for Model 3 age albumin Angiopoietin-2 bilirubin BUN chloride CO2 C-reactive protein Creatinine diastolic blood pressure Heart rate IL-6 IL-8 IP-10 lactic acid lymphocytes neutrophils PDL-1 Pentanoin-3 platelets Potassium Procalcitonin Pulse oxygen saturation respiratory rate sodium systolic blood pressure body temperature TREM-1 White blood cell count In one embodiment of Model 3, Figure 5A (No sepsis occurred within 24 hours) Figure 5B (No sepsis occurred within 24 hours) and Figure 5C The section on "(Sepsis occurring within 24 hours)" provides examples of SHAP values ​​for three different high-risk sepsis patients. The order of importance of the input features differs for these three patients.

[0165] Figure 5A This image shows a snapshot of Patient 1's SHAP values ​​for the listed input characteristics. The patient has both extremely high positive values ​​and extremely low negative values. Abnormal blood pressure readings, elevated creatinine, and multiple protein biomarkers (such as IL-8 and TREM-1) all contribute to an increased sepsis immune score. However, other biomarkers (including pentamericin-3, IL-6, and CRP), along with respiratory rate and normal oxygen saturation, contribute to a decreased risk score.

[0166] Figure 5BA snapshot of the SHAP values ​​for Patient 2 for the listed input characteristics was displayed. Normal respiratory rate and traditional laboratory indicators such as BUN, bilirubin, and creatinine had a significant negative impact on the patient's sepsis immune score, exceeding the effects of elevated IL-8, IP-10, and advanced age.

[0167] Figure 5C A snapshot of Patient 3's SHAP values ​​for the listed input characteristics is shown. This patient had the highest SHAP value for pentameric-3, as well as abnormal values ​​for other infection-related biomarkers and organ dysfunction indicators (platelets, pulse oxygen saturation).

[0168] Exemplary results of the diagnostic and prognostic performance of Model 3 are as follows: VE Model 4 The following are exemplary input features for Model 4, in no particular order.

[0169] Table 12: Examples of input features for Model 4 age albumin Angiopoietin-1 Angiopoietin-2 bilirubin BUN chloride CO2 C-reactive protein Creatinine diastolic blood pressure E-selectin FLT3 ligands Chemokines GCSF GMCSF Granulase B Heart rate IFN-α IFN-γ IL-10 IL-15 IL1-β IL1-RA IL-2 IL-4 IL-6 IL-7 IL-8 IP-10 lactic acid Leptin lymphocytes MCP1 MIP1-α MIP1-β MIP3-α neutrophils NGAL PDL-1 Pentanoin-3 platelets Potassium Procalcitonin Pulse oxygen saturation respiratory rate sodium systolic blood pressure body temperature TGF-α Thrombomodulin Organization Factor TNF-α TRAIL TREM-1 VCAM-1 VEG-F White blood cell count In one embodiment of Model 4 Figure 6A (No sepsis occurred within 24 hours) Figure 6B (Sepsis occurred within 24 hours) and Figure 6C The section "(No sepsis within 24 hours)" provides examples of SHAP values ​​for three different high-risk sepsis patients. The input features are ordered differently for each patient.

[0170] Figure 6A A snapshot of Patient 1's SHAP values ​​for the listed input characteristics is shown. Abnormal blood pressure indicates cardiovascular dysfunction, making the patient's blood pressure readings their most important SHAP values.

[0171] Figure 6B A snapshot of the SHAP values ​​for patient 2 for the listed input characteristics is shown. Low platelet count is the most important characteristic for this patient, as it indicates a coagulation disorder.

[0172] Figure 6C A snapshot of the SHAP values ​​for patient 3 for the listed input characteristics is shown. Creatinine, IL-8, and granzyme B have positive SHAP values, while platelet count, blood pressure, and other characteristics are normal, negatively impacting the patient's risk of sepsis.

[0173] Exemplary results of the diagnostic and prognostic performance of Model 4 are as follows: VF Model 5 The following are exemplary input features for Model 5, in no particular order.

[0174] Table 13: Examples of input features for Model 5 age Angiopoietin-2 BUN calcium chloride CO2 Creatinine diastolic blood pressure female glucose Heart rate Hematocrit hemoglobin IL-8 IP-10 Pentanoin-3 Potassium Pulse oxygen saturation respiratory rate sodium systolic blood pressure body temperature In one embodiment of Model 5 Figure 7A (Sepsis occurred within 24 hours) Figure 7B (No sepsis occurred within 24 hours) and Figure 7C The section "(No sepsis within 24 hours)" provides examples of SHAP values ​​for three different high-risk sepsis patients. The order of importance of the input features differs among these three patients.

[0175] Figure 7A A snapshot of Patient 1's SHAP values ​​for the listed input features is shown. Abnormal blood pressure (indicating cardiovascular organ dysfunction) is the feature that has the greatest impact on the patient's sepsis risk score.

[0176] Figure 7B A snapshot of Patient 2's SHAP values ​​for the listed input features is shown. Most of the patient's features have negative SHAP values, but the single most significant influence is their IL-8 value.

[0177] Figure 7C A snapshot of patient 3's SHAP values ​​for the listed input characteristics is shown. Abnormal pulse oxygen saturation and IL-8 had significant positive SHAP values, while most other laboratory indicators and biomarkers had negative effects on their sepsis risk score.

[0178] Exemplary results of the diagnostic and prognostic performance of Model 5 are as follows: VI. Examples of Input Features Specific embodiments of this disclosure provide risk stratification related to long-term and short-term prognostic clinical outcomes based on the analysis of one or more input features of any one or more of the features in Tables 8-13.

[0179] In various embodiments, an individual's sepsis risk index is provided by one or more input features from any one or more of Tables 8-13. In a particular embodiment, an individual's sepsis risk index is provided by input features 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 of Model 1A in Table 8. In a particular embodiment, an individual's sepsis risk index is provided by input features 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, or 23 of Model 1B in Table 9. In a specific embodiment, an individual's sepsis risk index is provided by input features 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, or 27 of Model 2 in Table 10. In a specific embodiment, an individual's sepsis risk index is provided by input features 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, or 29 of Model 3 in Table 11. In a specific embodiment, an individual's sepsis risk index is provided by one of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, or 58 input features of Model 4 in Table 12. In a particular embodiment, an individual’s sepsis risk index is provided by one of the following 22 input features of Model 5 in Table 13: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21 or 22.

[0180] In a specific embodiment, at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 input features of Model 1A are used as indicators of an individual's sepsis risk. In a specific embodiment, at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, or 23 input features of Model 1B are used as indicators of an individual's sepsis risk. In a specific embodiment, at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, or 27 input features of Model 2 are used as indicators of an individual's sepsis risk. In a specific embodiment, at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, or 29 input features of Model 3 are used as indicators of an individual's sepsis risk. In a specific embodiment, at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, or 58 input features of Model 4 are used as indicators of an individual's sepsis risk. In a specific embodiment, at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21 or 22 input features of Model 5 are used as indicators of an individual's sepsis risk.

[0181] In a specific embodiment, no more than 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 input features of Model 1A are used as an individual's sepsis risk indicator. In a specific embodiment, no more than 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, or 23 input features of Model 1B are used as an individual's sepsis risk indicator. In a specific embodiment, no more than 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, or 27 input features of Model 2 are used as an individual's sepsis risk indicator. In a specific embodiment, no more than 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28 or 29 input features of Model 3 are used as indicators of an individual's sepsis risk. In a specific embodiment, no more than 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, or 58 input features of Model 4 are used as indicators of sepsis risk for an individual. In a specific embodiment, no more than 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21 or 22 input features of Model 5 are used as indicators of an individual's sepsis risk.

[0182] In various embodiments, a majority of the input features of models 1A, 1B, 2, 3, 4, or 5 are used as an indicator of an individual's sepsis risk. In some embodiments, at least 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, or 95% of the input features of models 1A, 1B, 2, 3, 4, or 5 are used as an indicator of an individual's sepsis risk.

[0183] In a particular embodiment, at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 of the input features of Model 1A are used to determine an individual's risk of developing sepsis or to determine if they have sepsis. In certain cases, the order of importance of the input features may vary from patient to patient.

[0184] In a particular embodiment, at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, or 23 of the input features of Model 1B are used to determine an individual's risk of developing sepsis or to determine if they have sepsis. In certain cases, the order of importance of the input features may vary from patient to patient.

[0185] In a particular embodiment, at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, or 27 of the input features of Model 2 are used to determine an individual's risk of developing sepsis or to determine if they have sepsis. In certain cases, the order of importance of the input features may vary from patient to patient.

[0186] In a particular embodiment, at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, or 29 of the input features of Model 3 are used to determine an individual's risk of developing sepsis or to determine if they have sepsis. In certain cases, the order of importance of the input features may vary from patient to patient.

[0187] In a specific embodiment, at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, or 58 of the input features of Model 4 are used to determine an individual's risk of developing sepsis or to determine if they have sepsis. In certain cases, the order of importance of the input features may vary from patient to patient.

[0188] In a particular embodiment, at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 of the input features of Model 5 are used to determine an individual's risk of developing sepsis or to determine if they have sepsis. In certain cases, the order of importance of the input features may vary from patient to patient.

[0189] In any of the models described herein, patient age may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, patient albumin level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, patient angiopoietin-1 level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, patient angiopoietin-2 level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, patient bilirubin level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, patient physiological sex (e.g., female) may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, patient blood urea nitrogen (BUN) level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's chloride level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. Similarly, a patient's CO2 level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's C-reactive protein level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's creatinine level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's diastolic blood pressure level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's E-selectin level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. Finally, in any of the models described herein, a patient's FLT-3 ligand level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's chemokine levels may or may not be the most important input feature for determining the risk of developing sepsis or for determining the presence of sepsis. Similarly, in any of the models described herein, a patient's GCSF levels may or may not be the most important input feature for determining the risk of developing sepsis or for determining the presence of sepsis.In any of the models described herein, a patient's GMCSF level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. Similarly, in any of the models described herein, a patient's granzyme B level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's glucose level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's heart rate may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's hematocrit level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's hemoglobin level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. Finally, in any of the models described herein, a patient's IFN-α level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's IFN-γ level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's IL-10 level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's IL-15 level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's IL-1β level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's IL-1RA level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's IL-2 level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's IL-4 level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's IL-6 level may or may not be the most important input feature for determining the risk of developing sepsis or for determining the presence of sepsis. Similarly, in any of the models described herein, a patient's IL-7 level may or may not be the most important input feature for determining the risk of developing sepsis or for determining the presence of sepsis.In any of the models described herein, a patient's IL-8 level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's IL-10 level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's IP-10 level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's lactate level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's leptin level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's lymphocyte level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's MCP1 level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's MIP1-α level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's MIP1-β level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's MIP3-α level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's neutrophil level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's NGAL level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's platelet level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's PD-L1 level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's pentameric-3 level may or may not be the most important input feature for determining the risk of developing sepsis or for determining the presence of sepsis. Similarly, in any of the models described herein, a patient's potassium level may or may not be the most important input feature for determining the risk of developing sepsis or for determining the presence of sepsis.In any of the models described herein, a patient's platelet level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's procalcitonin level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's pulse oxygen saturation may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's respiratory rate may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's sodium level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's systolic blood pressure may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's body temperature may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's TGF-α level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. Similarly, in any of the models described herein, a patient's thrombomodulin level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's tissue factor level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's TNF-α level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's TRAIL level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's TREM-1 level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. Finally, in any of the models described herein, a patient's VCAM-1 level may or may not be the most important input feature for determining the risk of developing sepsis or the presence of sepsis. In any of the models described herein, a patient's VEG-F level may or may not be the most important input feature for determining the risk of developing sepsis or for determining the presence of sepsis. Similarly, in any of the models described herein, a patient's white blood cell count may or may not be the most important input feature for determining the risk of developing sepsis or for determining the presence of sepsis.

[0190] In various embodiments, the methods of this disclosure generate sepsis risk indicators, such as predictive scores, based on at least one input feature of any one or more models in Tables 8-13. These predictive scores indicate, for example, the probability that a patient will have or develop sepsis over a predetermined time span. In some embodiments, a particular input feature is more important than one or more other input features in the predictive score.

[0191] In a specific embodiment of Model 1A, age is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 1A, albumin level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 1A, bilirubin level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 1A, BUN level is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 1A, chloride level is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 1A, CO2 is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 1A, C-reactive protein level is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 1A, creatinine level is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 1A, diastolic blood pressure is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 1A, heart rate is more important in the predicted score than any of the other 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features.In a specific embodiment of Model 1A, lactate level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 1A, lymphocyte level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 1A, neutrophil level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 1A, platelet level is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 1A, potassium level is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 1A, procalcitonin level is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 1A, pulse oxygen saturation is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 1A, respiratory rate is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 1A, sodium level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 1A, systolic blood pressure is more important in the predicted score than any of the other 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features.In a specific embodiment of Model 1A, body temperature is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. Similarly, in a specific embodiment of Model 1A, white blood cell count is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features.

[0192] In a specific embodiment of Model 1B, age is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 other input features. In a specific embodiment of Model 1B, albumin level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 other input features. In a specific embodiment of Model 1B, bilirubin level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 other input features. In a specific embodiment of Model 1B, BUN level is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 other input features. In a specific embodiment of Model 1B, chloride level is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 other input features. In a specific embodiment of Model 1B, CO2 is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 other input features. In a specific embodiment of Model 1B, C-reactive protein level is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 other input features. In a specific embodiment of Model 1B, creatinine level is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 other input features. In a specific embodiment of Model 1B, diastolic blood pressure is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 other input features. In a specific embodiment of Model 1B, heart rate is more important in the predicted score than any of the other 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 other input features.In a specific embodiment of Model 1B, IL-6 levels are more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 other input features. In a specific embodiment of Model 1B, lactate levels are more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 other input features. In a specific embodiment of Model 1B, lymphocyte levels are more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 other input features. In a specific embodiment of Model 1B, neutrophil levels are more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 other input features. In a specific embodiment of Model 1B, platelet levels are more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 other input features. In a specific embodiment of Model 1B, potassium levels are more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 other input features. In a specific embodiment of Model 1B, procalcitonin level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 other input features. In a specific embodiment of Model 1B, pulse oxygen saturation is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 other input features. In a specific embodiment of Model 1B, respiratory rate is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 other input features. In a specific embodiment of Model 1B, sodium level is more important in the predicted score than any of the other 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 other input features.In a specific embodiment of Model 1B, systolic blood pressure is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 other input features. In a specific embodiment of Model 1B, body temperature is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 other input features. In a specific embodiment of Model 1B, white blood cell count is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or 22 other input features.

[0193] In a specific embodiment of Model 2, age is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features. In a specific embodiment of Model 2, albumin level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features. In a specific embodiment of Model 2, the angiopoietin-2 level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features. In a specific embodiment of Model 2, the bilirubin level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features. In a specific embodiment of Model 2, BUN level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features. Similarly, in a specific embodiment of Model 2, chloride level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features. In a specific embodiment of Model 2, CO2 is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features. In a specific embodiment of Model 2, C-reactive protein levels are more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features. In a specific embodiment of Model 2, creatinine level is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features.In a specific embodiment of Model 2, diastolic blood pressure is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features. Similarly, in a specific embodiment of Model 2, heart rate is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features. In a specific embodiment of Model 2, the IL-6 level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features. In a specific embodiment of Model 2, the IL-8 level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features. In a specific embodiment of Model 2, the IP-10 level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features. Similarly, in a specific embodiment of Model 2, the lactate level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features. In a specific embodiment of Model 2, lymphocyte level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features. In a specific embodiment of Model 2, neutrophil level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features. In a specific embodiment of Model 2, the pentameric protein-3 level is more important in the predicted score than any of the other 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features.In a specific embodiment of Model 2, platelet level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features. Similarly, in a specific embodiment of Model 2, potassium level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features. In a specific embodiment of Model 2, procalcitonin level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features. In a specific embodiment of Model 2, pulse oxygen saturation is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features. In a specific embodiment of Model 2, respiratory rate is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features. In a specific embodiment of Model 2, sodium level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features. In a specific embodiment of Model 2, systolic blood pressure is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features. In a specific embodiment of Model 2, body temperature is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features. In a specific embodiment of Model 2, white blood cell count is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, or 26 other input features.

[0194] In a specific embodiment of Model 3, age is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. In a specific embodiment of Model 3, albumin level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. In a specific embodiment of Model 3, the angiopoietin-2 level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. In a specific embodiment of Model 3, the bilirubin level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. In a specific embodiment of Model 3, BUN level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. Similarly, in a specific embodiment of Model 3, chloride level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. In a specific embodiment of Model 3, CO2 is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. In a specific embodiment of Model 3, C-reactive protein levels are more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features.In a specific embodiment of Model 3, creatinine level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. In a specific embodiment of Model 3, diastolic blood pressure is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. In a specific embodiment of Model 3, heart rate is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. In a specific embodiment of Model 3, IL-6 level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. In a specific embodiment of Model 3, the IL-8 level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. In a specific embodiment of Model 3, the IP-10 level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. In a specific embodiment of Model 3, lactate level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. In a specific embodiment of Model 3, lymphocyte level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features.In a specific embodiment of Model 3, neutrophil levels are more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. In a specific embodiment of Model 3, PD-L1 levels are more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. In a specific embodiment of Model 3, pentamericin-3 levels are more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. Similarly, in a specific embodiment of Model 3, platelet levels are more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. In a specific embodiment of Model 3, potassium level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. In a specific embodiment of Model 3, procalcitonin level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. In a specific embodiment of Model 3, pulse oxygen saturation is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. In a specific embodiment of Model 3, respiratory rate is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features.In a specific embodiment of Model 3, sodium level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. In a specific embodiment of Model 3, systolic blood pressure is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. In a specific embodiment of Model 3, body temperature is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. In a specific embodiment of Model 3, TREM-1 level is more important in the predicted score than any of the following 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 other input features. In a specific embodiment of Model 3, white blood cell count is more important in the predicted score than any of the other input features, such as 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28.

[0195] In a specific embodiment of Model 4, age is more important in the predicted score than any of the following input features: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, albumin level is more important in the predicted score than any of the other input features, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, the angiopoietin-1 level is more important in the predicted score than any of the other input features, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, the angiopoietin-2 level is more important in the predicted score than any of the other input features, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57.In a specific embodiment of Model 4, bilirubin levels are more important than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57 other input features in the predicted score. In a specific embodiment of Model 4, the BUN level is more important than any of the other input features 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57 in the predicted score. In a specific embodiment of Model 4, chloride levels are more important than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57 other input features in the predicted score. In a specific embodiment of Model 4, the CO2 level is more important than any of the other input features 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57 in the predicted score.In a specific embodiment of Model 4, C-reactive protein levels are more important in the predicted score than any of the other input features, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, creatinine level is more important in the predicted score than any of the other input features, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, diastolic blood pressure is more important in the predicted score than any of the following input features: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, the E-selectin level is more important in the predicted score than any of the following input features: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57.In a specific embodiment of Model 4, the FLT3 ligand level is more important in the predicted score than any of the following other input features: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, the chemokine level is more important than any of the other input features 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57 in the predicted score. In a specific embodiment of Model 4, the GCSF level is more important than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57 other input features in the predicted score. In a specific embodiment of Model 4, the GMCSF level is more important than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57 other input features in the predicted score.In a specific embodiment of Model 4, granzyme B level is more important in the predicted score than any of the other input features 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, heart rate is more important than any of the following input features in the predicted score: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, the IFN-α level is more important than any of the other input features 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57 in the predicted score. In a specific embodiment of Model 4, the IFN-γ level is more important than any of the other input features 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57 in the predicted score.In a specific embodiment of Model 4, the IL-10 level is more important in the predicted score than any of the following other input features: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, the IL-15 level is more important in the predicted score than any of the following input features: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, the IL1-β level is more important in the predicted score than any of the following input features: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, the IL1-RA level is more important in the predicted score than any of the other input features 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57.In a specific embodiment of Model 4, the angiopoietin-2 level is more important in the predicted score than any of the other input features, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, the IL-4 level is more important in the predicted score than any of the following input features: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, the IL-6 level is more important in the predicted score than any of the following input features: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, the IL-7 level is more important in the predicted score than any of the other input features, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57.In a specific embodiment of Model 4, the IL-8 level is more important in the predicted score than any of the following input features: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, the IP-10 level is more important than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57 other input features in the predicted score. In a specific embodiment of Model 4, lactate level is more important in the predicted score than any of the following input features: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, leptin levels are more important than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57 other input features in the predicted score.In a specific embodiment of Model 4, lymphocyte levels are more important than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57 other input features in the predicted score. In a specific embodiment of Model 4, the MCP1 level is more important than any of the following input features in the predicted score: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, the MIP1-α level is more important in the predicted score than any of the following input features: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, the MIP1-β level is more important in the predicted score than any of the following input features: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57.In a specific embodiment of Model 4, the MIP3-α level is more important than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57 other input features in the predicted score. In a specific embodiment of Model 4, neutrophil levels are more important in the predicted score than any of the other input features, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, the NGAL level is more important than any of the following input features in the predicted score: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, the PD-L1 level is more important than any of the following input features in the predicted score: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57.In a specific embodiment of Model 4, the pentamericin-3 level is more important in the predicted score than any of the other input features, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, platelet levels are more important than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57 other input features in the predicted score. In a specific embodiment of Model 4, potassium level is more important in the predicted score than any of the other input features, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, the procalcitonin level is more important in the predicted score than any of the other input features 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57.In a specific embodiment of Model 4, pulse oxygen saturation is more important in the predicted score than any of the other input features, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, respiratory rate is more important in the predicted score than any of the other input features, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, sodium level is more important in the predicted score than any of the other input features, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, systolic blood pressure is more important in the predicted score than any of the other input features, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, body temperature is more important in the predicted score than any of the other input features, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57.In a specific embodiment of Model 4, the TGF-α level is more important in the predicted score than any of the following input features: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, the level of thrombomodulin is more important in the predicted score than any of the other input features, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, the level of the organization factor is more important than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57 other input features in the predicted score. In a specific embodiment of Model 4, the TNF-α level is more important in the predicted score than any of the other input features, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57.In a specific embodiment of Model 4, the TRAIL level is more important in the predicted score than any of the following input features: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57. In a specific embodiment of Model 4, the TREM-1 level is more important than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57 other input features in the predicted score. In a specific embodiment of Model 4, the VCAM-1 level is more important than any of the other input features 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57 in the predicted score. In a specific embodiment of Model 4, the VEGF level is more important than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57 other input features in the predicted score.In a specific embodiment of Model 4, white blood cell count is more important in the predicted score than any of the other input features, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, or 57.

[0196] In a specific embodiment of Model 5, age is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 5, angiogenesis-2 level is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 5, BUN level is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 5, calcium level is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 5, chloride level is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 5, CO2 is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 5, creatinine level is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 5, diastolic blood pressure is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 5, biological sex is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 5, glucose level is more important in the predicted score than any of the other 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features.In a specific embodiment of Model 5, heart rate is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 5, hematocrit is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 5, hemoglobin is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 5, the IL-8 level is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 5, the IP-10 level is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 5, the pentameric protein-3 level is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 5, potassium level is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 5, pulse oximetry is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 5, respiratory rate is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 5, sodium level is more important in the predicted score than any of the other 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features.In a specific embodiment of Model 5, systolic blood pressure is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features. In a specific embodiment of Model 5, body temperature is more important in the predicted score than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or 21 other input features.

[0197] VII. Examples of Treatment Methods In some embodiments, individuals identified as having sepsis or at risk of sepsis (including those at risk of sepsis for 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48 hours or longer) may be given a therapeutically effective amount of one or more therapies. In specific embodiments, the therapy includes one or more antibiotics, for example, administered orally (tablets, liquids, or capsules), by injection, drops, and / or topical application (cream or ointment) and / or intravenous injection. Treatment may be administered approximately over the following timeframes: for example, 1–24 hours, 1–12 hours, 1–6 hours, 6–24 hours, 6–18 hours, 6–12 hours, 12–24 hours, 12–18 hours, or 18–24 hours; 1–7 days, 1–6 days, 1–5 days, 1–4 days, 1–3 days, 1–2 days, 2–7 days, 2–6 days, 2–5 days, 2–4 days, 2–3 days, 3–7 days, 3–6 days, 3–5 days, 3–4 days, 4–7 days, 4–6 days, 4–5 days, 5–7 days, 5–6 days, or 6–7 days; or 1–4 weeks, 1–3 weeks, 1–2 weeks, 2–4 weeks, 2–3 weeks, or 3–4 weeks or more, depending on the type of infection and its response to antibiotics. In some embodiments, if the first treatment is ineffective or less effective than expected, a second treatment may be necessary, for example, using a different type of antibiotic.

[0198] In specific embodiments, when an individual requires antibiotics, one or more of the following drugs may be administered: (1) ceftriaxone (e.g., 1 to 2 grams intravenously or intramuscularly every 12 to 24 hours for adults; 50 to 75 mg / kg / day intravenously or intramuscularly every 12 to 24 hours for infants, children, and adolescents; and 50 mg / kg / dose intravenously or intramuscularly every 24 hours for newborns and preterm infants); (2) piperacillin-tazobactam (e.g., 4.5 grams intravenously every 6 hours for adults (4 grams of piperacillin and 0.5 grams of tazobactam); 80 mg / kg / dose of piperacillin component (90 mg / kg / dose of piperacillin; tazobactam) intravenously every 6 hours for infants, children, and adolescents; and 80 or 100 mg / kg / dose of piperacillin component (90 mg / kg / dose of piperacillin; tazobactam) intravenously every 6 hours for newborns); (3) cefepime (e.g., 2 grams intravenously every 8 hours for adults; 50 mg / kg / dose of piperacillin; tazobactam) intravenously every 8 hours for infants; (3) Children and adolescents: 50 mg / kg / dose intravenously every 8 hours (maximum dose: 2 g / dose); and newborns: 30 or 50 mg / kg / dose intravenously every 12 hours; (4) Ceftazidime (e.g., 2 g intravenously every 8 hours for adults and adolescents; 90 to 150 mg / kg / day intravenously every 8 hours for infants and children (maximum dose: 6 g / day); 50 mg / kg / dose intravenously every 8 or 12 hours for newborns); (5) Vancomycin (e.g., 20 to 35 mg / kg / dose intravenously for adults (maximum dose: 3000 mg / dose), followed by 15 to 20 mg / kg / dose intravenously every 8 to 12 hours; 60 to 70 mg / kg / day intravenously every 6 to 8 hours for children and adolescents; 60 to 80 mg / kg / day intravenously every 6 hours for infants); and / or (6) Ciprofloxacin (e.g., 600 mg intravenously every 12 hours for adults).

[0199] Embodiments of this disclosure include methods for treating sepsis, applicable to individuals who have sepsis, are suspected of having sepsis, or are at risk of sepsis within 1-48 hours, including individuals who, based on input characteristics of the individual, use a machine learning model to generate a predictive score indicating that the individual has sepsis or that the individual will develop sepsis within a predetermined time span (e.g., 1-48 hours).

[0200] Embodiments of this disclosure include a method in which a diagnostic output is generated based on a predictive score according to any of the methods described herein, the diagnostic output classifying a patient as having sepsis or at risk of developing sepsis within a predetermined time span; the method further includes administering a therapeutically effective amount of sepsis treatment to the patient. In a specific embodiment, based on the generated diagnostic output, the treatment includes at least one of antibiotics or intravenous infusion.

[0201] Embodiments of this disclosure include a method for treating a patient with sepsis who has a score based on multiple input features that indicates the probability of the patient having or developing sepsis based on any of the methods described herein, the treatment comprising administering an effective amount of one or more sepsis therapies to the patient.

[0202] Embodiments of this disclosure include a method that includes measuring multiple input features about individuals and / or samples from individuals in Tables 8, 9, 10, 11, 12 and / or 13, optionally wherein the measurements of at least a first input feature and at least a second input feature are sequential, and the measurement of the second input feature depends on the measurement result of the first input feature.

[0203] Embodiments of this disclosure include a method comprising measuring 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, or 13 of any one or more of Tables 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or more of the following input features about an individual: In a specific embodiment, the individual has one or more symptoms of sepsis. In some embodiments, the individual is suspected of having sepsis, is suspected of developing sepsis, or is at risk of developing sepsis. In some embodiments, the individual is suspected of developing sepsis within 24 hours of a measurement taken from a sample obtained from the individual.

[0204] Embodiments of this disclosure include methods for treating individuals with sepsis who have one or more specific values ​​of any input feature in any one or more of Tables 8, 9, 10, 11, 12, or 13 that indicate they have sepsis.

[0205] Embodiments of this disclosure include a method of treating an individual having one or more specific values ​​of any one or more input features of any one or more of any one or more of Tables 8, 9, 10, 11, 12, or 13, wherein the treatment comprises administering a therapeutically effective amount of one or more antibiotics.

[0206] VIII. Computer-implemented systems Figure 8 This is a block diagram of a computer system according to various embodiments. The computer system 800 can be implemented in various ways (e.g., as described above). Figure 2 An example of an implementation of the method described in (200).

[0207] In one or more examples, computer system 800 may include a bus 802 or other communication mechanism for transmitting information, and a processor 804 coupled to the bus 802 for processing information. In various embodiments, computer system 800 may also include memory coupled to the bus 802, which may be random access memory (RAM) 806 or other dynamic storage device, for determining instructions to be executed by processor 804. The memory may also be used to store temporary variables or other intermediate information during instruction execution by processor 804. In various embodiments, computer system 800 may also include read-only memory (ROM) 808 or other static storage device coupled to the bus 802 for storing static information and instructions of processor 804. Storage device 810 (e.g., a magnetic disk or optical disk) may be coupled to the bus 802 for storing information and instructions.

[0208] In various embodiments, computer system 800 may be coupled to display 812, such as a cathode ray tube (CRT), liquid crystal display (LCD), or light-emitting diode (LED), via bus 802 for displaying information to a computer user. Input device 814 (including alphanumeric keys and other buttons) may be coupled to bus 802 for transmitting information and command selections to processor 804. Another type of user input device is cursor control device 816, such as a mouse, joystick, trackball, gesture input device, gaze-based input device, or cursor direction keys, for transmitting directional information and command selections to processor 804 and for controlling cursor movement on display 812. This input device 814 typically has two degrees of freedom on two axes, namely a first axis (e.g., the x-axis) and a second axis (e.g., the y-axis), allowing the device to specify a position in a plane. However, it should be understood that input devices 814 that allow three-dimensional (e.g., x, y, and z) cursor movement are also considered here.

[0209] According to certain implementations of this teaching, computer system 800 can provide results in response to processor 804 executing one or more sequences of instructions contained in RAM 806. These instructions may be read into RAM 806 from other computer-readable media or computer-readable storage media (e.g., storage device 810). Executing the sequence of instructions contained in RAM 806 causes processor 804 to perform the processes described herein. Note that this teaching may be implemented using hardwired circuitry instead of or in combination with software instructions. Therefore, implementations of this teaching are not limited to any particular combination of hardware circuitry and software.

[0210] As used herein, the term "computer-readable medium" (e.g., data storage device, data storage apparatus, storage device, etc.) or "computer-readable storage medium" refers to any medium that participates in providing execution instructions to processor 804. Such media can take many forms, including but not limited to non-volatile media, volatile media, and transmission media. Examples of non-volatile media include, but are not limited to, optical discs, solid-state drives, and magnetic disks (e.g., storage device 810). Examples of volatile media include, but are not limited to, dynamic memory (e.g., RAM 806). Examples of transmission media include, but are not limited to, coaxial cables, copper wires, and optical fibers (including conductors constituting bus 802).

[0211] Common forms of computer-readable media include, for example, floppy disks, flexible disks, hard disks, magnetic tapes or any other magnetic media, CD-ROMs, any other optical media, punched cards, paper tapes, any other physical media with a perforated pattern, RAM, PROMs and EPROMs, flash memory EPROMs, any other memory chips or tapes, or any other tangible media that a computer can read.

[0212] In addition to computer-readable media, instructions or data may also be provided as signals on a transmission medium included in a communication device or system to provide one or more sequences of instructions for execution to a processor 804 of a computer system 800. For example, a communication device may include a transceiver having signals indicating instructions and data. These instructions and data are configured to cause one or more processors to perform the functions outlined in this disclosure. Representative examples of data communication transmission connections include, but are not limited to, telephone modem connections, wide area networks (WANs), local area networks (LANs), infrared data connections, NFC connections, optical communication connections, etc.

[0213] It should be understood that the methods, flowcharts, diagrams, and accompanying disclosures described herein can be implemented using a computer system 800 as a standalone device or on a distributed network (such as a cloud computing network) sharing computer processing resources.

[0214] The methods described herein can be implemented in various ways depending on the specific application. For example, these methods can be implemented using hardware, firmware, software, or any combination thereof. For hardware implementation, the processing unit can be implemented in one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), processors, controllers, microcontrollers, microprocessors, electronic devices, other electronic units designed to perform the functions described herein, or combinations thereof.

[0215] In various embodiments, the methods of this teaching can be implemented as firmware and / or software programs, as well as applications written in traditional programming languages ​​such as C, C++, and Python. If implemented as firmware and / or software, the embodiments described herein can be implemented on a non-transitory computer-readable medium storing programs for causing a computer to perform the methods described above. It should be understood that the various engines described herein can be deployed on a computer system (e.g., computer system 800), wherein processor 804 performs the analysis and determination provided by these engines based on instructions provided by one or more or combinations of memory components RAM 806, ROM 808, or storage device 810, and user input provided via input device 814.

[0216] IX. Other Considerations Any headings and / or subheadings between sections and subsections in this document are for readability purposes only and do not imply that features from different sections and subsections cannot be combined. Therefore, no individual embodiment is described in any section or subsection.

[0217] While this teaching has been described in conjunction with various embodiments, it is not intended to limit the teaching to these embodiments. Rather, the teaching encompasses various alternatives, modifications, and equivalents that will be understood by those skilled in the art. This specification provides preferred exemplary embodiments and is not intended to limit the scope, applicability, or configuration of this disclosure. Rather, the description of preferred exemplary embodiments will provide those skilled in the art with effective guidance for implementing the various embodiments.

[0218] It should be understood that various changes can be made to the function and arrangement of the elements without departing from the spirit and scope of the appended claims. Therefore, such modifications and variations should be considered within the scope of the appended claims. Furthermore, the terminology and expressions used are for descriptive purposes only and not for limitation, and their use is not intended to exclude equivalents of the shown and described features or any part thereof, but rather to acknowledge that various modifications can be made within the scope of the claimed invention.

[0219] In describing various embodiments, the specification may describe methods and / or processes as a specific sequence of steps. However, if a method or process does not depend on the specific sequence of steps described herein, the method or process should not be limited to the specific sequence of steps described, and those skilled in the art will readily understand that these sequences may vary but remain within the spirit and scope of the various embodiments.

[0220] Some embodiments of this disclosure include a system comprising one or more data processors. In some embodiments, the system includes a non-transitory computer-readable storage medium containing instructions that, when executed on the one or more data processors, cause the one or more data processors to perform part or all of the one or more methods and / or one or more processes disclosed herein. Some embodiments of this disclosure include a computer program product tangibly embodied in a non-transitory machine-readable storage medium including instructions configured to cause one or more data processors to perform part or all of the one or more methods and / or one or more processes disclosed herein.

[0221] Specific details are provided in this specification to aid in understanding the various embodiments. However, it should be understood that these embodiments can be implemented even without these specific details. For example, circuits, systems, networks, processes, and other components may be represented in the form of block diagrams to avoid obscuring the embodiments with unnecessary details. In other cases, well-known circuits, processes, algorithms, structures, and techniques may omit unnecessary details to avoid obscuring the embodiments.

[0222] Example 1. A method for generating sepsis risk indicators, the method comprising: Receive at least one input feature corresponding to the patient; The at least one input feature is analyzed using a machine learning model to generate a predictive score based on the at least one input feature, the predictive score indicating the probability that the patient has or develops sepsis within a predetermined time span; wherein the at least one input feature is associated with sepsis.

[0223] 2. The method as described in Example 1, wherein the at least one input feature includes clinical parameters.

[0224] 3. The method as described in Example 1 or 2, wherein the at least one input feature includes a biomarker.

[0225] 4. The method as described in any one of Examples 1-3 further includes: generating a diagnostic output based on the predicted score.

[0226] 5. The method as described in any one of Examples 1-4, wherein the clinical parameters and / or the biomarkers are selected from the input feature groups listed in Table 8.

[0227] 6. The method as described in any one of Examples 1-4, wherein the clinical parameters and / or the biomarkers are selected from the input feature groups listed in Table 9.

[0228] 7. The method as described in any one of Examples 1-4, wherein the clinical parameters and / or the biomarkers are selected from the input feature groups listed in Table 10.

[0229] 8. The method as described in any one of Examples 1-4, wherein the clinical parameters and / or the biomarkers are selected from the input feature groups listed in Table 11.

[0230] 9. The method as described in any one of Examples 1-4, wherein the clinical parameters and / or the biomarkers are selected from the input feature groups listed in Table 12.

[0231] 10. The method as described in any one of Examples 1-4, wherein the clinical parameters and / or the biomarkers are selected from the input feature groups listed in Table 13.

[0232] 11. The method as described in any one of Examples 5-10, wherein the input feature groups listed in any one of Tables 1-6 are listed according to their relative importance to the predicted score.

[0233] 12. The method of any one of Examples 1-11, wherein the relative importance of the predicted score is determined based on feature ranking, wherein the feature ranking is determined by ranking the input features based on the average SHAP value, wherein a high average SHAP value indicates high relative importance of the predicted score, and a low SHAP value indicates low relative importance of the predicted score.

[0234] 13. The method as described in Example 12, wherein the average SHAP value of the input features in any of the input feature groups listed in Tables 8-13 is calculated by the following steps: Determine the absolute values ​​of all SHAP values ​​for a given population of the input features; Add the absolute values ​​of all SHAP values ​​to obtain the sum of the input features; and Divide the sum by the number of populations in the given population of the input features.

[0235] 14. The method of any one of Examples 1-13, wherein the at least one input feature includes procalcitonin, wherein procalcitonin concentration values ​​are divided into two groups, wherein the relative importance of the first group of procalcitonin concentration values ​​is lower than that of the second group of procalcitonin concentration values.

[0236] 15. The method as described in Example 14, wherein the threshold for dividing the first group of procalcitonin concentration values ​​and the second group of procalcitonin concentration values ​​is 2.5 log 10 pm / mL.

[0237] 16. The method as described in Example 15, wherein the relative importance of procalcitonin to the predicted score in each of the first and second groups is the median SHAP value of the procalcitonin concentration values ​​within the corresponding group.

[0238] 17. The method of any one of Examples 1-13, wherein the at least one input feature includes procalcitonin, wherein the procalcitonin concentration values ​​are divided into at least three groups, wherein the relative importance of the first group of procalcitonin concentration values ​​is lower than that of the second group of procalcitonin concentration values, and wherein the relative importance of the third group of procalcitonin concentration values ​​is higher than that of the first group of procalcitonin concentration values ​​and the second group of procalcitonin concentration values.

[0239] 18. The method as described in Example 17, wherein the first dividing threshold between the first group of procalcitonin concentration values ​​and the second group of procalcitonin concentration values ​​is 1.7 log10 pg / mL, and wherein the second dividing threshold between the second group of procalcitonin concentration values ​​and the third group of procalcitonin concentration values ​​is 3.4 log10 pg / mL.

[0240] 19. The method as described in Example 18, wherein the relative importance of procalcitonin to the predicted score in each of the first, second, and third groups is the median SHAP value of the procalcitonin concentration values ​​within the corresponding group.

[0241] 20. The method as described in any one of Examples 1-19, wherein the input feature groups listed in Table 1 include: Glasgow Coma Scale score, FiO2, procalcitonin, systolic blood pressure, and platelet count.

[0242] 21. The method as described in Example 20, wherein the Glasgow Coma Scale score is of higher relative importance than FiO2, FiO2 is of higher relative importance than procalcitonin, procalcitonin is of higher relative importance than systolic blood pressure, and systolic blood pressure is of higher relative importance than platelet count.

[0243] 22. The method as described in any one of Examples 1-19, wherein the input feature groups listed in Table 1 include: Glasgow Coma Scale score, FiO2, procalcitonin, systolic blood pressure, platelet count, bilirubin, blood urea nitrogen, respiratory rate, diastolic blood pressure, and albumin.

[0244] 23. The method as described in any one of Examples 1-19, wherein the input feature groups listed in Table 1 include: Glasgow Coma Scale score, FiO2, procalcitonin, systolic blood pressure, platelet count, bilirubin, blood urea nitrogen, respiratory rate, diastolic blood pressure, albumin, creatinine, age, blood oxygen saturation, lactate, and C-reactive protein.

[0245] 24. The method as described in any one of Examples 1-19, wherein the input feature groups listed in Table 1 include: Glasgow Coma Scale score, FiO2, procalcitonin, systolic blood pressure, platelet count, bilirubin, blood urea nitrogen, respiratory rate, diastolic blood pressure, albumin, creatinine, age, oxygen saturation, lactate, C-reactive protein, potassium, neutrophil count, body temperature, total carbon dioxide, white blood cell count, lymphocyte count, sodium, heart rate, and chloride.

[0246] 25. The method of any one of Examples 1-19, wherein the input feature set includes procalcitonin and C-reactive protein.

[0247] 26. The method as described in Example 25, wherein one or both of procalcitonin and C-reactive protein are replaced by one or more other biomarkers listed in Table 2 based on correlation values ​​associated with Table 3.

[0248] 27. The method as described in Example 26, wherein the correlation value of one or more other biomarkers replacing one or both of procalcitonin and C-reactive protein is less than XX.

[0249] 28. The method as described in any one of Examples 1-27, further comprising: Determine that the predicted score belongs to one of four categories; and A diagnostic output is generated based on the category to which the predicted score belongs. The diagnostic output includes: the probability that the patient has or develops sepsis within the predetermined time span is low, moderate, high, or very high.

[0250] 29. The method as described in Example 28, wherein the predicted score ranges from 0 to 1.

[0251] 30. The method as described in Example 28 or 29, wherein the four categories include a low category, a medium category, a high category, and an extremely high category, and wherein the low category is designated as having a predicted score below a first threshold, the medium category is designated as having a predicted score between the first threshold and a second threshold, the high category is designated as having a predicted score between the second threshold and a third threshold, and the extremely high category is designated as having a predicted score above the third threshold.

[0252] 31. The method as described in Example 30, wherein the predicted score is below the first threshold of 0.122, and the diagnostic output indicates that the probability of the patient having or developing sepsis within the predetermined time span is low.

[0253] 32. The method as described in Example 30, wherein the predicted score is equal to or higher than the first threshold of 0.122 and lower than the second threshold of 0.306, and the diagnostic output indicates that the patient has a moderate probability of having or developing sepsis within the predetermined time span.

[0254] 33. The method as described in Example 30, wherein the predicted score is equal to or higher than the second threshold of 0.306 and lower than the third threshold of 0.872, and the diagnostic output indicates that the patient has a high probability of having or developing sepsis within the predetermined time span.

[0255] 34. The method as described in Example 30, wherein the predicted score is equal to or higher than the third threshold of 0.872, and the diagnostic output indicates that the patient has a very high probability of having or developing sepsis within the predetermined time span.

[0256] 35. The method as described in any one of Examples 1-34, further comprising: The predicted score is calibrated using a calibration model to generate a calibration score; and Risk categories are generated as diagnostic output based on the calibration score.

[0257] 36. The method as described in Example 35, wherein generating the diagnostic output includes: The calibration score was determined to belong to one of four risk categories; and Based on the category to which the calibration score belongs, the risk category is generated, which includes: the probability that the patient has or develops sepsis within the predetermined time span is low, moderate, high, or very high.

[0258] 37. The method as described in Example 36, wherein the calibration score ranges from 0 to 1.

[0259] 38. The method of any one of Examples 35-37, wherein the risk category can be selected from a group consisting of a low category, a medium category, a high category, and an extremely high category, wherein the low category is designated as the calibration score being below a first threshold, the medium category is designated as the calibration score being between the first threshold and a second threshold, the high category is designated as the calibration score being between the second threshold and a third threshold, and the extremely high category is designated as the calibration score being above the third threshold.

[0260] 39. The method as described in Example 38, wherein the calibration score is below the first threshold of 0.122, and the risk category indicates that the patient has a low probability of having or developing sepsis within the predetermined time span.

[0261] 40. The method as described in Example 38, wherein the calibration score is higher than the first threshold of 0.122 and lower than the second threshold of 0.306, and the risk category indicates that the patient has a moderate probability of having or developing sepsis within the predetermined time span.

[0262] 41. The method as described in Example 38, wherein the calibration score is higher than the second threshold of 0.306 and lower than the third threshold of 0.872, and the risk category indicates that the patient has a high probability of having or developing sepsis within the predetermined time span.

[0263] 42. The method as described in Example 38, wherein the calibration score is higher than the third threshold of 0.872, and the risk category indicates that the patient has an extremely high probability of having or developing sepsis within the predetermined time span.

[0264] 43. The method as described in any one of Examples 1-42, wherein the predetermined time span does not exceed or exceeds 12 hours, 18 hours, 24 hours, 36 hours, 48 ​​hours or 72 hours.

[0265] 44. The method as described in any one of Examples 1-43, further comprising: At least one machine learning model is trained using training data, wherein the training data includes multiple biomarkers and health profiles of multiple patients and multiple subject diagnoses of the multiple patients.

[0266] 45. The method as described in any one of Examples 4-44, wherein the diagnostic output is a label automatically derived from the predicted score generated by the at least one machine learning model.

[0267] 46. ​​The method of any one of Examples 4-45, wherein the diagnostic output is determined by a physician based on the predicted score generated by the at least one machine learning model.

[0268] 47. The method of any one of Examples 4-46, wherein the analysis is performed without a value for one of a plurality of clinical parameters and / or a value for one of a plurality of biomarkers to generate the predictive score, the method further comprising: generating the diagnostic output based on the predictive score.

[0269] 48. The method as described in any one of Examples 1-47, further comprising: Identify and impute the value of one of the input features, where the input feature is missing a value associated with the input feature.

[0270] 49. The method of any one of Embodiments 1-47, wherein one of the input features has a missing value, the method further comprises: The trigger logic decision module is used to determine whether to perform the analysis to generate the predicted score in the absence of the missing value of the input feature.

[0271] 50. The method of any one of Examples 4-47, wherein one of the input features has a missing value, the method further comprises: The missing values ​​of the input feature are imputed to generate imputed values ​​for the missing input features; Analyze data including the interpolated values ​​to generate the predictive score, which indicates the probability that the patient has or will develop sepsis within a predetermined time span; and A diagnostic output is generated based on the predicted score.

[0272] 51. The method of any one of Examples 48-50, wherein the imputation of the missing values ​​is performed using a pre-trained template trained with training data, the training data including input features including multiple clinical parameters and multiple biomarkers of multiple patients, and multiple subject diagnoses of the multiple patients.

[0273] 52. The method of any one of Examples 48-51, wherein the input feature having the missing value is specified as a feature whose value is measured or obtained outside a specified time window in which data for generating the predicted score can be accepted.

[0274] 53. The method as described in Example 52, wherein the specified time window is the same time period for the first group of input features, and is different for at least two input features in the second group of input features.

[0275] 54. The method as described in any one of Examples 1-53, wherein the input features include clinical parameters having values ​​measured within a specified time window.

[0276] 55. The method of any one of Examples 1-54, wherein the input feature includes a biomarker having a measurement value associated with the biomarker acquired within a specified time window.

[0277] 56. The method of any one of Examples 1-55, wherein the input features include clinical parameters, which include at least one of demographic measurements, patient assessments, vital signs, hematological laboratory values, and chemical laboratory values.

[0278] 57. The method as described in any one of Examples 48-50, further comprising: Identify and imput biomarkers that are missing values ​​associated with other biomarkers among the plurality of biomarkers.

[0279] 58. The method of any one of Examples 48-57, wherein the input feature missing a value associated with the input feature includes: biomarkers missing values ​​associated with the biomarkers from the plurality of biomarkers, and / or clinical parameters missing values ​​associated with the clinical parameters from the plurality of clinical parameters.

[0280] 59. The method as described in any one of Examples 1-58, wherein the input features include at least 20 clinical parameters and at least two biomarkers.

[0281] 60. The method of any one of Examples 5-10, wherein the input feature group listed in any one of Tables 1-6 includes at least one biomarker feature and at least one clinical parameter feature, wherein the values ​​of one or more biomarker features are divided into two groups, and wherein the values ​​of one or more clinical parameter features are divided into two groups.

[0282] 61. The method as described in Example 60, wherein the biomarker features include at least one of procalcitonin, C-reactive protein, interleukin-6, pentamin-3, interleukin-8, interleukin-1 receptor antagonist protein, and vascular cell adhesion molecule 1, wherein the relative importance of a first set of concentration values ​​for each biomarker feature is lower than that of a second set of concentration values ​​for the same biomarker feature.

[0283] 62. The method as described in Example 60 or 61, wherein the clinical parameter includes at least one of platelets, systolic blood pressure, diastolic blood pressure, albumin and blood oxygen saturation, and wherein the relative importance of a first set of values ​​for each clinical parameter feature is higher than that of a second set of concentration values ​​for the clinical parameter feature.

[0284] 63. The method as described in Example 60 or 61, wherein the clinical parameter includes at least one of respiratory rate, blood urea nitrogen, bilirubin and creatinine, and wherein the relative importance of a first set of values ​​for each clinical parameter feature is lower than that of a second set of concentration values ​​for the clinical parameter feature.

[0285] 64. The method as described in Example 60, wherein the biomarker features include procalcitonin, wherein the relative importance of a first group of procalcitonin concentration values ​​is lower than that of a second group of procalcitonin concentration values.

[0286] 65. The method as described in Example 60, wherein the biomarker features include C-reactive protein, wherein the relative importance of the first group of C-reactive protein concentration values ​​is lower than that of the second group of C-reactive protein concentration values.

[0287] 66. The method as described in Example 60, wherein the biomarker features include interleukin-6, wherein the relative importance of the first group of interleukin-6 concentration values ​​is lower than that of the second group of interleukin-6 concentration values.

[0288] 67. The method as described in Example 60, wherein the biomarker features include pentamericin 3, wherein the relative importance of the first group of pentamericin 3 concentration values ​​is lower than that of the second group of pentamericin 3 concentration values.

[0289] 68. The method as described in Example 60, wherein the biomarker features include interleukin-8, wherein the relative importance of the first group of interleukin-8 concentration values ​​is lower than that of the second group of interleukin-8 concentration values.

[0290] 69. The method as described in Example 60, wherein the biomarker features include interleukin-1 receptor antagonist proteins, wherein the relative importance of the concentration values ​​of the first group of interleukin-1 receptor antagonist proteins is lower than that of the concentration values ​​of the second group of interleukin-1 receptor antagonist proteins.

[0291] 70. The method as described in Example 60, wherein the biomarker features include vascular cell adhesion molecules 1, wherein the relative importance of the concentration values ​​of the first group of vascular cell adhesion molecules 1 is lower than that of the concentration values ​​of the second group of vascular cell adhesion molecules 1.

[0292] 71. The method as described in Example 60, wherein the biomarker features include procalcitonin and C-reactive protein, wherein the relative importance of the first group of procalcitonin concentration values ​​is lower than that of the second group of procalcitonin concentration values, and wherein the relative importance of the first group of C-reactive protein concentration values ​​is lower than that of the second group of C-reactive protein concentration values.

[0293] 72. The method as described in Example 60, wherein the biomarker features include procalcitonin and interleukin-6, wherein the relative importance of the first group of procalcitonin concentration values ​​is lower than that of the second group of procalcitonin concentration values, and wherein the relative importance of the first group of interleukin-6 concentration values ​​is lower than that of the second group of interleukin-6 concentration values.

[0294] 73. The method as described in Example 60, wherein the biomarker features include interleukin-6 and C-reactive protein, wherein the relative importance of the first group of interleukin-6 concentration values ​​is lower than that of the second group of interleukin-6 concentration values, and wherein the relative importance of the first group of C-reactive protein concentration values ​​is lower than that of the second group of C-reactive protein concentration values.

[0295] 74. The method as described in Example 60, wherein the biomarker features include procalcitonin, C-reactive protein, and interleukin-6, wherein the relative importance of the first group of procalcitonin concentration values ​​is lower than that of the second group of procalcitonin concentration values, wherein the relative importance of the first group of C-reactive protein concentration values ​​is lower than that of the second group of C-reactive protein concentration values, and wherein the relative importance of the first group of interleukin-6 concentration values ​​is lower than that of the second group of interleukin-6 concentration values.

[0296] 75. The method as described in Example 60, wherein the biomarker features include pentamericin 3 and interleukin-1 receptor antagonist protein, wherein the relative importance of the concentration values ​​of the first group of pentamericin 3 is lower than that of the second group of pentamericin 3, and wherein the relative importance of the concentration values ​​of the first group of interleukin-1 receptor antagonist protein is lower than that of the second group of interleukin-1 receptor antagonist protein.

[0297] 76. The method as described in Example 60, wherein the biomarker features include pentamericin 3 and interleukin 8, wherein the relative importance of the first group of pentamericin 3 concentration values ​​is lower than that of the second group of pentamericin 3 concentration values, and wherein the relative importance of the first group of interleukin 8 concentration values ​​is lower than that of the second group of interleukin 8 concentration values.

[0298] 77. The method as described in Example 60, wherein the biomarker features include procalcitonin and interleukin-8, wherein the relative importance of the first group of procalcitonin concentration values ​​is lower than that of the second group of procalcitonin concentration values, and wherein the relative importance of the first group of interleukin-8 concentration values ​​is lower than that of the second group of interleukin-8 concentration values.

[0299] 78. The method as described in Example 60, wherein the biomarker features include procalcitonin and pentaprotein 3, wherein the relative importance of the first group of procalcitonin concentration values ​​is lower than that of the second group of procalcitonin concentration values, and wherein the relative importance of the first group of pentaprotein 3 concentration values ​​is lower than that of the second group of pentaprotein 3 concentration values.

[0300] 79. The method as described in Example 60, wherein the biomarker features include pentamericin 3 and interleukin 6, wherein the relative importance of the first group of pentamericin 3 concentration values ​​is lower than that of the second group of pentamericin 3 concentration values, and wherein the relative importance of the first group of interleukin 6 concentration values ​​is lower than that of the second group of interleukin 6 concentration values.

[0301] 80. The method as described in Example 60, wherein the clinical parameter features include platelets, wherein the relative importance of the first group of platelet values ​​is higher than that of the second group of platelet values.

[0302] 81. The method as described in Example 60, wherein the clinical parameter features include systolic blood pressure, wherein the relative importance of a first group of systolic blood pressure values ​​is higher than that of a second group of systolic blood pressure values.

[0303] 82. The method as described in Example 60, wherein the clinical parameter features include diastolic blood pressure, wherein the relative importance of a first group of diastolic blood pressure values ​​is higher than that of a second group of diastolic blood pressure values.

[0304] 83. The method as described in Example 60, wherein the clinical parameter features include blood oxygen saturation, wherein the relative importance of the first group of blood oxygen saturation values ​​is higher than that of the second group of blood oxygen saturation values.

[0305] 84. The method as described in Example 60, wherein the clinical parameter features include albumin, wherein the relative importance of a first group of albumin values ​​is higher than that of a second group of albumin values.

[0306] 85. The method as described in Example 60, wherein the clinical parameter features include respiratory rate, wherein the relative importance of a first group of respiratory rate values ​​is lower than that of a second group of respiratory rate values.

[0307] 86. The method as described in Example 60, wherein the clinical parameter features include blood urea nitrogen, wherein the relative importance of the first group of blood urea nitrogen values ​​is lower than that of the second group of blood urea nitrogen values.

[0308] 87. The method as described in Example 60, wherein the clinical parameter features include bilirubin, wherein the relative importance of the first group of bilirubin values ​​is lower than that of the second group of bilirubin values.

[0309] 88. The method as described in Example 60, wherein the clinical parameter features include creatinine, wherein the relative importance of the first group of creatinine values ​​is lower than that of the second group of creatinine values.

[0310] 89. The method as described in Example 60, wherein the clinical parameter features include creatinine and blood urea nitrogen, wherein the relative importance of the first group of creatinine values ​​is lower than that of the second group of creatinine values, and wherein the relative importance of the first group of blood urea nitrogen values ​​is lower than that of the second group of blood urea nitrogen values.

[0311] 90. The method as described in Example 60, wherein the clinical parameter characteristics include platelets, creatinine, and blood urea nitrogen, wherein the relative importance of a first group of platelet values ​​is higher than that of a second group of platelet values, wherein the relative importance of a first group of creatinine values ​​is lower than that of a second group of creatinine values, and wherein the relative importance of a first group of blood urea nitrogen values ​​is lower than that of a second group of blood urea nitrogen values.

[0312] 91. The method as described in Example 60, wherein the clinical parameter features include respiratory rate, creatinine and blood urea nitrogen, wherein the relative importance of the first group of respiratory rate values ​​is lower than that of the second group of respiratory rate values, wherein the relative importance of the first group of creatinine values ​​is lower than that of the second group of creatinine values, and wherein the relative importance of the first group of blood urea nitrogen values ​​is lower than that of the second group of blood urea nitrogen values.

[0313] 92. The method as described in Example 60, wherein the clinical parameter features include platelets and respiratory rate, wherein the relative importance of a first group of platelet values ​​is higher than that of a second group of platelet values, and wherein the relative importance of a first group of respiratory rate values ​​is lower than that of a second group of respiratory rate values.

[0314] 93. The method as described in Example 60, wherein the clinical parameter characteristics include platelets and systolic blood pressure, wherein the relative importance of the first group of platelet values ​​is higher than that of the second group of platelet values, and wherein the relative importance of the first group of systolic blood pressure values ​​is lower than that of the second group of systolic blood pressure values.

[0315] 94. The method as described in Example 60, wherein the clinical parameter features include platelets, respiratory rate and systolic blood pressure, wherein the relative importance of the first group of platelet values ​​is higher than that of the second group of platelet values, wherein the relative importance of the first group of respiratory rate values ​​is lower than that of the second group of respiratory rate values, and wherein the relative importance of the first group of systolic blood pressure values ​​is higher than that of the second group of systolic blood pressure values.

[0316] 95. The method as described in Example 60, wherein the relative importance of one or more features in each of the first and second groups to the predicted score is the median SHAP value of the feature values ​​within the respective group.

[0317] 96. The method as described in Example 95, wherein the relative importance of one or more biomarker features in each of the first and second groups to the predicted score is the median SHAP value of the biomarker feature values ​​within the respective group.

[0318] 97. The method as described in Example 95, wherein the relative importance of one or more clinical parameter features in each of the first and second groups to the predicted score is the median SHAP value of the clinical parameter feature values ​​within the corresponding group.

[0319] 98. The method as described in Example 60, wherein the relative importance of one or more features in each of the first and second groups to the predicted score is the average SHAP value of the feature values ​​within the respective group.

[0320] 99. The method as described in Example 98, wherein the relative importance of one or more biomarker features in each of the first and second groups to the predicted score is the average SHAP value of the biomarker feature values ​​within the respective group.

[0321] 100. The method as described in Example 98, wherein the relative importance of one or more clinical parameter features in each of the first and second groups to the predicted score is the average SHAP value of the clinical parameter feature values ​​within the corresponding group.

[0322] 101. The method as described in any one of Examples 61-100, wherein the threshold for dividing the first group of procalcitonin concentration values ​​and the second group of procalcitonin concentration values ​​is 2.24 log 10 pg / mL.

[0323] 102. The method as described in any one of Examples 61-100, wherein the threshold for dividing the first group of pentameric protein 3 concentration values ​​and the second group of pentameric protein 3 concentration values ​​is 3.98 log10 pg / mL.

[0324] 103. The method as described in any one of Examples 61-100, wherein the threshold for dividing the first group of interleukin-8 concentration values ​​and the second group of interleukin-8 concentration values ​​is 1.41 log10 pg / mL.

[0325] 104. The method as described in any one of Examples 61-100, wherein the threshold for dividing the concentration values ​​of the first group of interleukin-1 receptor antagonist proteins and the second group of interleukin-1 receptor antagonist proteins is 3.48 log10 pg / mL.

[0326] 105. The method as described in any one of Examples 61-100, wherein the threshold for dividing the first group of interleukin-6 concentration values ​​and the second group of interleukin-6 concentration values ​​is 2.07 log10 pg / mL.

[0327] 106. The method as described in any one of Examples 61-100, wherein the threshold for dividing the concentration values ​​of the first group of C-reactive protein and the second group of C-reactive protein is 7.63 log10 pg / mL.

[0328] 107. The method as described in any one of Examples 61-100, wherein the threshold for dividing the concentration values ​​of the first group of vascular cell adhesion molecules 1 and the second group of vascular cell adhesion molecules 1 is 6.17 log10 pg / mL.

[0329] 108. The method as described in any one of Examples 61-100, wherein the threshold for dividing the first group of platelet concentration values ​​and the second group of platelet concentration values ​​is 162 × 10⁹ / L.

[0330] 109. The method as described in any one of Examples 61-100, wherein the threshold for dividing the first group of systolic blood pressure values ​​and the second group of systolic blood pressure values ​​is 101 mmHg.

[0331] 110. The method as described in any one of Examples 61-100, wherein the threshold for dividing the first group of respiratory rate values ​​and the second group of respiratory rate values ​​is 101 breaths per minute.

[0332] 111. The method as described in any one of Examples 61-100, wherein the threshold for dividing the first group of blood urea nitrogen values ​​and the second group of blood urea nitrogen values ​​is 24 mg / dL.

[0333] 112. The method as described in any one of Examples 61-100, wherein the threshold for dividing the first group of bilirubin values ​​and the second group of bilirubin values ​​is 1.19 mg / dL.

[0334] 113. The method as described in any one of Examples 61-100, wherein the threshold for dividing the first group of creatinine values ​​and the second group of creatinine values ​​is 1.56 mg / dL.

[0335] 114. The method as described in any one of Examples 61-100, wherein the threshold for dividing the first group of diastolic blood pressure values ​​and the second group of diastolic blood pressure values ​​is 54.5 mmHg.

[0336] 115. The method as described in any one of Examples 61-100, wherein the threshold for dividing the first group of blood oxygen saturation values ​​and the second group of blood oxygen saturation values ​​is 91%.

[0337] 116. The method of any one of Examples 5-10, wherein the input feature group listed in any one of Tables 1-6 includes at least one biomarker feature and at least one clinical parameter feature, wherein the values ​​of one or more biomarker features are divided into at least three groups, and wherein the values ​​of one or more clinical parameter features are divided into at least three groups.

[0338] 117. The method as described in Example 116, wherein the biomarker features include at least one of procalcitonin, C-reactive protein, interleukin-6, pentamin-3, interleukin-8, interleukin-1 receptor antagonist protein, and vascular cell adhesion molecule 1, wherein the relative importance of a first set of concentration values ​​for each biomarker feature is lower than that of a second set of concentration values ​​for the same biomarker feature, and wherein the relative importance of a third set of concentration values ​​for the same biomarker feature is higher than that of the first set of concentration values ​​and the second set of concentration values.

[0339] 118. The method as described in Example 116, wherein the clinical parameter includes at least one of platelets, systolic blood pressure, diastolic blood pressure, albumin and pulse oxygen saturation, and wherein the relative importance of a first set of values ​​for each clinical parameter feature is higher than that of a second set of concentration values ​​for the clinical parameter feature, and wherein the relative importance of a third set of concentration values ​​for the clinical parameter feature is lower than that of the first set of clinical parameter values ​​and the second set of clinical parameter values.

[0340] 119. The method as described in Example 118, wherein the clinical parameters include at least one of respiratory rate, blood urea nitrogen, bilirubin and creatinine, and wherein the relative importance of a first set of values ​​for each clinical parameter feature is lower than that of a second set of concentration values ​​for the clinical parameter feature, and wherein the relative importance of a third set of values ​​for each clinical parameter feature is higher than that of the first set of clinical parameter feature values ​​and the second set of clinical parameter feature values.

[0341] 120. The method as described in Example 116, wherein the biomarker features include procalcitonin, wherein the relative importance of a first group of procalcitonin concentration values ​​is lower than that of a second group of procalcitonin concentration values, and wherein the relative importance of a third group of procalcitonin concentration values ​​is higher than that of the first group of procalcitonin concentration values ​​and the second group of procalcitonin concentration values.

[0342] 121. The method as described in Example 116, wherein the biomarker features include C-reactive protein, wherein the relative importance of a first group of C-reactive protein concentration values ​​is lower than that of a second group of C-reactive protein concentration values, and wherein the relative importance of a third group of C-reactive protein concentration values ​​is higher than that of the first group of C-reactive protein concentration values ​​and the second group of C-reactive protein concentration values.

[0343] 122. The method as described in Example 116, wherein the biomarker features include interleukin-6, wherein the relative importance of a first group of interleukin-6 concentration values ​​is lower than that of a second group of interleukin-6 concentration values, and wherein the relative importance of a third group of interleukin-6 concentration values ​​is higher than that of the first group of interleukin-6 concentration values ​​and the second group of interleukin-6 concentration values.

[0344] 123. The method as described in Example 116, wherein the biomarker features include pentamericin 3, wherein the relative importance of the first group of pentamericin 3 concentration values ​​is lower than that of the second group of pentamericin 3 concentration values, and wherein the relative importance of the third group of pentamericin 3 concentration values ​​is higher than that of the first group of pentamericin 3 concentration values ​​and the second group of pentamericin 3 concentration values.

[0345] 124. The method as described in Example 116, wherein the biomarker features include interleukin-8, wherein the relative importance of a first group of interleukin-8 concentration values ​​is lower than that of a second group of interleukin-8 concentration values, and wherein the relative importance of a third group of interleukin-8 concentration values ​​is higher than that of the first group of interleukin-8 concentration values ​​and the second group of interleukin-8 concentration values.

[0346] 125. The method as described in Example 116, wherein the biomarker features include interleukin-1 receptor antagonist proteins, wherein the relative importance of the concentration values ​​of the first group of interleukin-1 receptor antagonist proteins is lower than that of the second group of interleukin-1 receptor antagonist proteins, and wherein the relative importance of the concentration values ​​of the third group of interleukin-1 receptor antagonist proteins is higher than that of the first group of interleukin-1 receptor antagonist proteins and the second group of interleukin-1 receptor antagonist proteins.

[0347] 126. The method as described in Example 116, wherein the biomarker features include vascular cell adhesion molecules 1, wherein the relative importance of the concentration values ​​of the first group of vascular cell adhesion molecules 1 is lower than that of the second group of vascular cell adhesion molecules 1, and wherein the relative importance of the concentration values ​​of the third group of vascular cell adhesion molecules 1 is higher than that of the first group of vascular cell adhesion molecules 1 and the second group of vascular cell adhesion molecules 1.

[0348] 127. The method as described in Example 116, wherein the biomarker features include procalcitonin and C-reactive protein, wherein the relative importance of a first group of procalcitonin concentration values ​​is lower than that of a second group of procalcitonin concentration values, and wherein the relative importance of a third group of procalcitonin concentration values ​​is higher than that of the first group of procalcitonin concentration values ​​and the second group of procalcitonin concentration values, and wherein the relative importance of a first group of C-reactive protein concentration values ​​is lower than that of the second group of C-reactive protein concentration values, and wherein the relative importance of a third group of C-reactive protein concentration values ​​is higher than that of the first group of C-reactive protein concentration values ​​and the second group of C-reactive protein concentration values.

[0349] 128. The method as described in Example 116, wherein the biomarker features include procalcitonin and interleukin-6, wherein the relative importance of a first group of procalcitonin concentration values ​​is lower than that of a second group of procalcitonin concentration values, and wherein the relative importance of a third group of procalcitonin concentration values ​​is higher than that of the first group of procalcitonin concentration values ​​and the second group of procalcitonin concentration values, and wherein the relative importance of a first group of interleukin-6 concentration values ​​is lower than that of the second group of interleukin-6 concentration values, and wherein the relative importance of a third group of interleukin-6 concentration values ​​is higher than that of the first group of interleukin-6 concentration values ​​and the second group of interleukin-6 concentration values.

[0350] 129. The method as described in Example 116, wherein the biomarker features include C-reactive protein and interleukin-6, wherein the relative importance of the concentration values ​​of the first group of C-reactive protein is lower than that of the second group of C-reactive protein, and wherein the relative importance of the concentration values ​​of the third group of C-reactive protein is higher than that of the first group of C-reactive protein and the second group of C-reactive protein, and wherein the relative importance of the concentration values ​​of the first group of interleukin-6 is lower than that of the second group of interleukin-6, and wherein the relative importance of the concentration values ​​of the third group of interleukin-6 is higher than that of the first group of interleukin-6 and the second group of interleukin-6.

[0351] 130. The method as described in Example 116, wherein the biomarker features include procalcitonin, C-reactive protein, and interleukin-6, wherein the relative importance of a first group of procalcitonin concentration values ​​is lower than that of a second group of procalcitonin concentration values, and wherein the relative importance of a third group of procalcitonin concentration values ​​is higher than that of the first and second groups of procalcitonin concentration values; wherein the relative importance of a first group of C-reactive protein concentration values ​​is lower than that of the second group of C-reactive protein concentration values, and wherein the relative importance of a third group of C-reactive protein concentration values ​​is higher than that of the first and second groups of C-reactive protein concentration values; wherein the relative importance of a first group of interleukin-6 concentration values ​​is lower than that of the second group of interleukin-6 concentration values, and wherein the relative importance of a third group of interleukin-6 concentration values ​​is higher than that of the first and second groups of interleukin-6 concentration values.

[0352] 131. The method as described in Example 116, wherein the biomarker features include pentaprotein 3 and interleukin-1 receptor antagonist protein, wherein the relative importance of the concentration values ​​of the first group of pentaprotein 3 is lower than that of the second group of pentaprotein 3, and wherein the relative importance of the concentration values ​​of the third group of pentaprotein 3 is higher than that of the first group of pentaprotein 3 and the second group of pentaprotein 3, and wherein the relative importance of the concentration values ​​of the first group of interleukin-1 receptor antagonist protein is lower than that of the second group of interleukin-1 receptor antagonist protein, and wherein the relative importance of the concentration values ​​of the third group of interleukin-1 receptor antagonist protein is higher than that of the first group of interleukin-1 receptor antagonist protein and the second group of interleukin-1 receptor antagonist protein.

[0353] 132. The method as described in Example 116, wherein the biomarker features include pentaprotein 3 and interleukin 8, wherein the relative importance of the first group of pentaprotein 3 concentration values ​​is lower than that of the second group of pentaprotein 3 concentration values, and wherein the relative importance of the third group of pentaprotein 3 concentration values ​​is higher than that of the first group of pentaprotein 3 concentration values ​​and the second group of pentaprotein 3 concentration values, and wherein the relative importance of the first group of interleukin 8 concentration values ​​is lower than that of the second group of interleukin 8 concentration values, and wherein the relative importance of the third group of interleukin 8 concentration values ​​is higher than that of the first group of interleukin 8 concentration values ​​and the second group of interleukin 8 concentration values.

[0354] 133. The method as described in Example 116, wherein the biomarker features include procalcitonin and interleukin-8, wherein the relative importance of the first group of procalcitonin concentration values ​​is lower than that of the second group of procalcitonin concentration values, and wherein the relative importance of the third group of procalcitonin concentration values ​​is higher than that of the first group of procalcitonin concentration values ​​and the second group of procalcitonin concentration values, wherein the relative importance of the first group of interleukin-8 concentration values ​​is lower than that of the second group of interleukin-8 concentration values, and wherein the relative importance of the third group of interleukin-8 concentration values ​​is higher than that of the first group of interleukin-8 concentration values ​​and the second group of interleukin-8 concentration values.

[0355] 134. The method as described in Example 116, wherein the biomarker features include procalcitonin and pentamericin 3, wherein the relative importance of the first group of procalcitonin concentration values ​​is lower than that of the second group of procalcitonin concentration values, and wherein the relative importance of the third group of procalcitonin concentration values ​​is higher than that of the first group of procalcitonin concentration values ​​and the second group of procalcitonin concentration values, and wherein the relative importance of the first group of pentamericin 3 concentration values ​​is lower than that of the second group of pentamericin 3 concentration values, and wherein the relative importance of the third group of pentamericin 3 concentration values ​​is higher than that of the first group of pentamericin 3 concentration values ​​and the second group of pentamericin 3 concentration values.

[0356] 135. The method as described in Example 116, wherein the biomarker features include pentamericin 3 and interleukin 6, wherein the relative importance of the first group of pentamericin 3 concentration values ​​is lower than that of the second group of pentamericin 3 concentration values, and wherein the relative importance of the third group of pentamericin 3 concentration values ​​is higher than that of the first group of pentamericin 3 concentration values ​​and the second group of pentamericin 3 concentration values, and wherein the relative importance of the first group of interleukin 6 concentration values ​​is lower than that of the second group of interleukin 6 concentration values, and wherein the relative importance of the third group of interleukin 6 concentration values ​​is higher than that of the first group of interleukin 6 concentration values ​​and the second group of interleukin 6 concentration values.

[0357] 136. The method as described in Example 116, wherein the clinical parameter features include platelets, wherein the relative importance of a first group of platelet values ​​is higher than that of a second group of platelet values, and wherein the relative importance of a third group of platelet values ​​is lower than that of the first group of platelet values ​​and the second group of platelet values.

[0358] 137. The method as described in Example 116, wherein the clinical parameter features include systolic blood pressure, wherein the relative importance of a first group of systolic blood pressure values ​​is higher than that of a second group of systolic blood pressure values, and wherein the relative importance of a third group of systolic blood pressure values ​​is lower than that of the first group of systolic blood pressure values ​​and the second group of systolic blood pressure values.

[0359] 138. The method as described in Example 116, wherein the clinical parameter features include diastolic blood pressure, wherein the relative importance of a first group of diastolic blood pressure values ​​is higher than that of a second group of diastolic blood pressure values, and wherein the relative importance of a third group of diastolic blood pressure values ​​is lower than that of the first group of diastolic blood pressure values ​​and the second group of diastolic blood pressure values.

[0360] 139. The method as described in Example 116, wherein the clinical parameter features include pulse oxygen saturation, wherein the relative importance of a first group of pulse oxygen saturation values ​​is higher than that of a second group of pulse oxygen saturation values, and wherein the relative importance of a third group of pulse oxygen saturation values ​​is lower than that of the first group of pulse oxygen saturation values ​​and the second group of pulse oxygen saturation values.

[0361] 140. The method as described in Example 116, wherein the clinical parameter features include albumin, wherein the relative importance of a first group of albumin values ​​is higher than that of a second group of albumin values, and wherein the relative importance of a third group of albumin values ​​is lower than that of the first group of albumin values ​​and the second group of albumin values.

[0362] 141. The method as described in Example 116, wherein the clinical parameter features include respiratory rate, wherein the relative importance of a first group of respiratory rate values ​​is higher than that of a second group of respiratory rate values, and wherein the relative importance of a third group of respiratory rate values ​​is lower than that of the first group of respiratory rate values ​​and the second group of respiratory rate values.

[0363] 142. The method as described in Example 116, wherein the clinical parameter features include blood urea nitrogen, wherein the relative importance of a first group of blood urea nitrogen values ​​is higher than that of a second group of blood urea nitrogen values, and wherein the relative importance of a third group of blood urea nitrogen values ​​is lower than that of the first group of blood urea nitrogen values ​​and the second group of blood urea nitrogen values.

[0364] 143. The method as described in Example 116, wherein the clinical parameter features include bilirubin, wherein the relative importance of a first group of bilirubin values ​​is higher than that of a second group of bilirubin values, and wherein the relative importance of a third group of bilirubin values ​​is lower than that of the first group of bilirubin values ​​and the second group of bilirubin values.

[0365] 144. The method as described in Example 116, wherein the clinical parameter features include creatinine, wherein the relative importance of a first group of creatinine values ​​is higher than that of a second group of creatinine values, and wherein the relative importance of a third group of creatinine values ​​is lower than that of the first group of creatinine values ​​and the second group of creatinine values.

[0366] 145. The method as described in Example 116, wherein the clinical parameter characteristics include creatinine and blood urea nitrogen, wherein the relative importance of a first group of creatinine values ​​is higher than that of a second group of creatinine values, and wherein the relative importance of a third group of creatinine values ​​is lower than that of the first group of creatinine values ​​and the second group of creatinine values, and wherein the relative importance of a first group of blood urea nitrogen values ​​is higher than that of the second group of blood urea nitrogen values, and wherein the relative importance of a third group of blood urea nitrogen values ​​is lower than that of the first group of blood urea nitrogen values ​​and the second group of blood urea nitrogen values.

[0367] 146. The method as described in Example 116, wherein the clinical parameter characteristics include platelets, creatinine, and blood urea nitrogen, wherein the relative importance of a first group of platelet values ​​is higher than that of a second group of platelet values, and wherein the relative importance of a third group of platelet values ​​is lower than that of the first and second groups of platelet values; wherein the relative importance of a first group of creatinine values ​​is higher than that of the second group of creatinine values, and wherein the relative importance of a third group of creatinine values ​​is lower than that of the first and second groups of creatinine values; wherein the relative importance of a first group of blood urea nitrogen values ​​is higher than that of the second group of blood urea nitrogen values, and wherein the relative importance of a third group of blood urea nitrogen values ​​is lower than that of the first and second groups of blood urea nitrogen values.

[0368] 147. The method as described in Example 116, wherein the clinical parameter features include respiratory rate, creatinine, and blood urea nitrogen, wherein the relative importance of a first group of respiratory rate values ​​is higher than that of a second group of respiratory rate values, and wherein the relative importance of a third group of respiratory rate values ​​is lower than that of the first group of respiratory rate values ​​and the second group of respiratory rate values; wherein the relative importance of a first group of creatinine values ​​is higher than that of a second group of creatinine values, and wherein the relative importance of a third group of creatinine values ​​is lower than that of the first group of creatinine values ​​and the second group of creatinine values; wherein the relative importance of a first group of blood urea nitrogen values ​​is higher than that of a second group of blood urea nitrogen values, and wherein the relative importance of a third group of blood urea nitrogen values ​​is lower than that of the first group of blood urea nitrogen values ​​and the second group of blood urea nitrogen values.

[0369] 148. The method as described in Example 116, wherein the clinical parameter features include platelets and respiratory rate, wherein the relative importance of a first group of platelet values ​​is higher than that of a second group of platelet values, and wherein the relative importance of a third group of platelet values ​​is lower than that of the first group of platelet values ​​and the second group of platelet values, and wherein the relative importance of a first group of respiratory rate values ​​is higher than that of the second group of respiratory rate values, and wherein the relative importance of a third group of respiratory rate values ​​is lower than that of the first group of respiratory rate values ​​and the second group of respiratory rate values.

[0370] 149. The method as described in Example 116, wherein the clinical parameter characteristics include platelets and systolic blood pressure, wherein the relative importance of a first group of platelet values ​​is higher than that of a second group of platelet values, and wherein the relative importance of a third group of platelet values ​​is lower than that of the first group of platelet values ​​and the second group of platelet values, and wherein the relative importance of a first group of systolic blood pressure values ​​is higher than that of the second group of systolic blood pressure values, and wherein the relative importance of a third group of systolic blood pressure values ​​is lower than that of the first group of systolic blood pressure values ​​and the second group of systolic blood pressure values.

[0371] 150. The method as described in Example 116, wherein the clinical parameter characteristics include platelets, respiratory rate, and systolic blood pressure, wherein the relative importance of a first group of platelet values ​​is higher than that of a second group of platelet values, and wherein the relative importance of a third group of platelet values ​​is lower than that of the first and second groups of platelet values; wherein the relative importance of a first group of respiratory rate values ​​is higher than that of the second group of respiratory rate values, and wherein the relative importance of a third group of respiratory rate values ​​is lower than that of the first and second groups of respiratory rate values; wherein the relative importance of a first group of systolic blood pressure values ​​is higher than that of the second group of systolic blood pressure values, and wherein the relative importance of a third group of systolic blood pressure values ​​is lower than that of the first and second groups of systolic blood pressure values.

[0372] 151. The method as described in Example 116, wherein the relative importance of one or more features in each of the first, second, and third groups to the predicted score is the median SHAP value of the feature values ​​within the respective group.

[0373] 152. The method as described in Example 151, wherein the relative importance of one or more biomarker features in each of the first, second, and third groups to the predicted score is the median SHAP value of the biomarker feature values ​​within the respective group.

[0374] 153. The method as described in Example 151, wherein the relative importance of one or more clinical parameter features in each of the first, second, and third groups to the predicted score is the median SHAP value of the clinical parameter feature values ​​within the corresponding group.

[0375] 154. The method as described in Example 116, wherein the relative importance of one or more features in each of the first, second, and third groups to the predicted score is the average SHAP value of the feature values ​​within the respective group.

[0376] 155. The method as described in Example 154, wherein the relative importance of one or more biomarker features in each of the first, second, and third groups to the predicted score is the average SHAP value of the biomarker feature values ​​within the respective group.

[0377] 156. The method as described in Example 154, wherein the relative importance of one or more clinical parameter features in each of the first, second, and third groups to the predicted score is the average SHAP value of the clinical parameter feature values ​​within the corresponding group.

[0378] 157. The method as described in any one of Examples 116-156, wherein the threshold for dividing the first group of procalcitonin concentration values ​​and the second group of procalcitonin concentration values ​​is 2.05 log10 pg / mL, and wherein the second threshold for dividing the second group of procalcitonin concentration values ​​and the third group of procalcitonin concentration values ​​is 2.85 log10 pg / mL.

[0379] 158. The method as described in any one of Examples 116-156, wherein the dividing threshold between the first group of pentamericin 3 concentration values ​​and the second group of pentamericin 3 concentration values ​​is 3.61 log 10 pg / mL, and wherein the second dividing threshold between the second group of pentamericin 3 concentration values ​​and the third group of pentamericin 3 concentration values ​​is 3.98 log 10 pg / mL.

[0380] 159. The method as described in any one of Examples 116-156, wherein the dividing threshold between the first group of interleukin-8 concentration values ​​and the second group of interleukin-8 concentration values ​​is 1.36 log10 pg / mL, and wherein the second dividing threshold between the second group of interleukin-8 concentration values ​​and the third group of interleukin-8 concentration values ​​is 1.97 log10 pg / mL.

[0381] 160. The method as described in any one of Examples 116-156, wherein the threshold for dividing the concentration values ​​of the first group of interleukin-1 receptor antagonists and the second group of interleukin-1 receptor antagonists is 3.15 log10 pg / mL, and wherein the second threshold for dividing the concentration values ​​of the second group of interleukin-1 receptor antagonists and the third group of interleukin-1 receptor antagonists is 3.57 log10 pg / mL.

[0382] 161. The method as described in any one of Examples 116-156, wherein the threshold for dividing the first group of interleukin-6 concentration values ​​and the second group of interleukin-6 concentration values ​​is 3.15 log10 pg / mL, and wherein the second threshold for dividing the second group of interleukin-6 concentration values ​​and the third group of interleukin-6 concentration values ​​is 3.57 log10 pg / mL.

[0383] 162. The method as described in any one of Examples 116-156, wherein the threshold for dividing the first group of C-reactive protein concentration values ​​and the second group of C-reactive protein concentration values ​​is 7.31 log10 pg / mL, and wherein the second threshold for dividing the second group of C-reactive protein concentration values ​​and the third group of C-reactive protein concentration values ​​is 7.91 log10 pg / mL.

[0384] 163. The method as described in any one of Examples 116-156, wherein the threshold for dividing the concentration values ​​of the first group of vascular cell adhesion molecules 1 and the second group of vascular cell adhesion molecules 1 is 6.05 log 10 pg / mL, and wherein the second threshold for dividing the concentration values ​​of the second group of vascular cell adhesion molecules 1 and the third group of vascular cell adhesion molecules 1 is 6.17 log 10 pg / mL.

[0385] 164. The method as described in any one of Examples 116-156, wherein the threshold for dividing the first group of platelet concentration values ​​and the second group of platelet concentration values ​​is 139 10^9 / L, and wherein the second threshold for dividing the second group of platelet concentration values ​​and the third group of platelet concentration values ​​is 188 10^9 / L.

[0386] 165. The method as described in any one of Examples 116-156, wherein the dividing threshold between the first group of systolic blood pressure values ​​and the second group of systolic blood pressure values ​​is 91 mmHg, and the second dividing threshold between the second group of systolic blood pressure values ​​and the third group of systolic blood pressure values ​​is 111 mmHg.

[0387] 166. The method as described in any one of Examples 116-156, wherein the threshold for dividing the first group of respiratory rate values ​​and the second group of respiratory rate values ​​is 23.5 breaths per minute, and wherein the second threshold for dividing the second group of respiratory rate values ​​and the third group of respiratory rate values ​​is 28 breaths per minute.

[0388] 167. The method as described in any one of Examples 116-156, wherein the threshold for dividing the first group of blood urea nitrogen values ​​and the second group of blood urea nitrogen values ​​is 21 mg / dL, and wherein the second threshold for dividing the second group of blood urea nitrogen values ​​and the third group of blood urea nitrogen values ​​is 41 mg / dL.

[0389] 168. The method as described in any one of Examples 116-156, wherein the threshold for dividing the first group of bilirubin values ​​and the second group of bilirubin values ​​is 0.88 mg / dL, and wherein the second threshold for dividing the second group of bilirubin values ​​and the third group of bilirubin values ​​is 1.5 mg / dL.

[0390] 169. The method as described in any one of Examples 116-156, wherein the threshold for dividing the first group of creatinine values ​​and the second group of creatinine values ​​is 1.15 mg / dL, and wherein the second threshold for dividing the second group of creatinine values ​​and the third group of creatinine values ​​is 1.73 mg / dL.

[0391] 170. The method as described in any one of Examples 116-156, wherein the dividing threshold between the first group of diastolic blood pressure values ​​and the second group of diastolic blood pressure values ​​is 52 mmHg, and wherein the second dividing threshold between the second group of diastolic blood pressure values ​​and the third group of diastolic blood pressure values ​​is 66 mmHg.

[0392] 171. The method as described in any one of Examples 116-156, wherein the dividing threshold between the first group of blood oxygen saturation values ​​and the second group of blood oxygen saturation values ​​is 81 mmHg, and wherein the second dividing threshold between the second group of blood oxygen saturation values ​​and the third group of blood oxygen saturation values ​​is 92.5 mmHg.

[0393] 172. A method wherein the diagnostic output is generated based on the predicted score according to any one of Examples 4-171, classifying the patient as having sepsis or at risk of developing sepsis within a predetermined time span, the method further comprising: The individual was given a therapeutically effective dose of sepsis treatment.

[0394] 173. The method as described in Example 172, wherein, based on the generated diagnostic output, the treatment comprises at least one of an antibiotic or an intravenous infusion.

[0395] 174. A method for treating an individual with sepsis, the individual having a score based on multiple input features, the score indicating the probability that the patient has or will develop sepsis based on a method as described in any one of Examples 1-171, the treatment comprising administering an effective amount of one or more sepsis therapies to the individual.

[0396] 175. A method comprising measuring multiple input features of an individual and / or samples from an individual in Tables 8, 9, 10, 11, 12 or 13, optionally wherein the measurements of at least a first input feature and at least a second input feature are sequential, and the measurement of the second input feature depends on the measurement of the first input feature.

[0397] 176. The method as described in Example 175, comprising: measuring most of the input features in Table 8 concerning a plurality of input features of the individual.

[0398] 177. The method as described in Example 175 or 176, wherein the individual is suspected of having sepsis, is suspected of developing sepsis, or is at risk of developing sepsis, or wherein the individual has one or more symptoms of sepsis.

[0399] 178. The method as described in any one of Examples 175-177, wherein the individual is suspected of developing sepsis within 24 hours after the individual is measured.

[0400] 179. The method of any one of Examples 175-178, wherein the sample is blood, urine, tissue, skin, saliva, sputum and / or mucus.

[0401] 180. The method as described in any one of Examples 175-179, wherein the individual has an infection.

[0402] 181. The method as described in Example 180, wherein the infection is a viral infection, a bacterial infection, a fungal infection, or a prion infection.

[0403] 182. The method as described in Examples 180 or 181, wherein the infection is SARS-CoV-2, influenza, meningitis, pneumonia, tuberculosis, or hepatitis.

[0404] 183. The method as described in any one of Examples 175-182, wherein the individual is older than about 65 years and younger than about 1 year, has a weakened immune system, suffers from a chronic disease, has recently been seriously ill or hospitalized, and / or has a history of sepsis.

[0405] 184. The method as described in Example 183, wherein the chronic disease is diabetes, lung disease, cancer, heart disease, hepatitis, or kidney disease.

[0406] 185. A method comprising measuring 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, or 13 of any one or more of Tables 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or more of input features about an individual.

[0407] 186. The method as described in Example 185, wherein the individual has one or more symptoms of sepsis.

[0408] 187. The method as described in Example 185 or 186, wherein the individual is suspected of having sepsis, is suspected of developing sepsis, or is at risk of developing sepsis.

[0409] 188. The method as described in any one of Examples 185-187, wherein the individual is suspected of developing sepsis within 24 hours after a measurement is taken from a sample obtained from the individual.

[0410] 189. The method as described in any one of Examples 185-188, wherein the individual has an infection.

[0411] 190. The method as described in Example 189, wherein the infection is a viral infection, a bacterial infection, a fungal infection, or a prion infection.

[0412] 191. The method as described in Examples 189 or 190, wherein the infection is SARS-CoV-2, influenza, meningitis, pneumonia, tuberculosis, or hepatitis.

[0413] 192. The method as described in any one of Examples 185-191, wherein the individual is older than about 65 years and younger than about 1 year, has a weakened immune system, suffers from a chronic disease, has recently been seriously ill or hospitalized, and / or has a history of sepsis.

[0414] 193. The method as described in Example 192, wherein the chronic disease is diabetes, lung disease, cancer, heart disease, or kidney disease.

[0415] 194. The method of any one of Examples 1-171, wherein the plurality of input features includes at least one input feature of a first group, at least one input feature of a second group, and at least one input feature of a third group.

[0416] 195. The method as described in Example 194, The first group includes input features selected from a set of the following: age, sex at birth, race, ethnicity, patient history, current chief complaint or symptoms, neurological assessment, clinical decision support alerts, clinician or other chart records, diagnostic codes, surgery performed on the patient, current medications, interventions, patient care environment, medical imaging data or assessments, electrocardiogram, endoscopy, systolic blood pressure, diastolic blood pressure, body temperature, respiratory rate, heart rate, blood oxygen saturation, inhaled oxygen concentration and combinations thereof; The second group includes input features selected from a set of the following: white blood cell count, lymphocyte count, neutrophil count, platelet count, creatinine, blood urea nitrogen, potassium, chloride, total carbon dioxide, sodium, albumin, bilirubin, lactate, glucose, calcium, hematocrit, hemoglobin concentration, laboratory-derived biomarkers, clinical severity indicators and composite indicators, biopsy data, microbial test data and combinations thereof; and The third group includes input features selected from a set of the following: procalcitonin, C-reactive protein, protein biomarkers, genomic biomarkers, gene expression or transcriptome biomarkers, composite biomarker scores and combinations thereof.

[0417] 196. The method as described in Embodiment 194 or 195, wherein the first set includes input features selected from the group consisting of: Vital signs (e.g., body temperature, heart rate, blood pressure, respiratory rate, blood oxygen saturation); Demographic information (e.g., age, sex, race, ethnicity); Past medical history; Current chief complaint or symptoms; Neurological assessment (Glasgow Coma Scale, Four-Nutrient Assessment (FOUR)); Clinical decision support alerts; Clinician or other chart / graph records; Diagnostic codes (e.g., ICD-10); Surgery (including CPT codes); Current medication (including NDC code); Intervention measures; Patient care environment (ICU, ED, wards, outpatient department, etc.); Imaging examinations (e.g., X-ray, CT scan, MRI, fMRI, ultrasound); Electrocardiogram (e.g., EEG, EKG); Endoscopic examination; and The combination of the above items.

[0418] 197. The method as described in any one of Examples 194-196, wherein the second group comprises input features selected from the group consisting of: Laboratory tests; Laboratory-derived biomarkers (e.g., monocyte distribution width, Cytovale IntelliSep); Clinical severity indicators and comprehensive indicators (e.g., Sequential Organ Failure Assessment (SOFA) score, SOFA component scores, Charlson Comorbidity Index, Acute Physiology and Chronic Health Assessment (APACHE) I and II); Biopsy; Microbial testing (culture, PCR testing, antigen testing); and The combination of the above items.

[0419] 198. The method as described in any one of Examples 194-196, wherein the third group comprises input features selected from the group consisting of: Protein biomarkers; Genomic biomarkers; Gene expression or transcriptome biomarkers; Composite biomarker scoring; and The combination of the above items.

[0420] 199. The method as described in Example 198, wherein the protein biomarker is selected from the group consisting of: angiopoietin-1, angiopoietin-2, C-reactive protein, cystatin C, D-dimer, E-selectin, chemokine, FLT3 ligand, GCSF, GDF15, GMCSF, granzyme B, IFN-α, IFN-γ, IL1-β, IL1-RA, IL-2, IL-4, IL-6, IL-7, IL-8, IL-10. IL-15, IP-10, lactate dehydrogenase (LDH), lipopolysaccharide-binding protein (LBP), leptin, MCP1, MIP1-α, MIP1-β, MIP3-α, NGAL, islet stone protein, PDL-1, pentaprotein-3, procalcitonin, protein C, Sl00b, TGF-α, thrombomodulin, tissue factor, TNF-α, TRAIL, TREM-1, troponin, VCAM-1, VEG-F and combinations thereof.

[0421] 200. The method as described in Examples 198 or 199, wherein the gene expression or transcriptome biomarkers are selected from the group consisting of: CEACAM4, LAMP1, PLAC8, PLA2G7, CEACAM1, ZDHHC19, C9orf95, GNA15, BATF, C3AR1, KIAA1370, TGFBI, MTCH1, RPGRIP1, HLA-DPB1, nuclear factor-κB (NF-κB), heme oxygenase-1 (HO-1), FCGR2C, CRP, IL-6, IL-10, IL-12, TNF, LY6E, LAMP3, ISG15, USP18, RSAD2, IFI44L , NCOA7, CMPK2, IFI44, BATF2, GBP4, IFIH1, GMPR, CXCL10, IFI27, ISG15, CCL2, OAS3, P2RY14, GCH1, CD274, ID01, ARL4A, BPGM, TNFAIP3, TNFAIP6, IL1R 1. PROK2, RGS1, CXCR6, RTP4, TLR3, FZD5, KCNJ2, IRAK2, RGS16, CXCL8, OLR1, CCRL2, SPP1, ST3GAL4, CST7, IL1R2, IL4R, GALM, CTLA4, SOCS1, BCL2L1, TG M2, SLC39A8, CA2, PNP, GATA1, SLC1A5, RGS16, CD44, PDGFC, CSF2RA, CXCL1, CSF3R, SOCS3, IL18R1, PIM1, LEPR, IL1B, CSF2RB, ITGB3, STAT1, FAS, EFNA1 , IER3, EETS2, SLC2A3, BCL2A1, PFKFB3, CEBPB, BIRC2, ATF3, TUBB2A, G0S2, MERTK, BIK, QSOX1, PYGL, TGFA, P4HA2, LDHA, IRS2, TSPO, HGF, GADD45A, TGF BR3, CD38, PRF1, FASLG, TIMP3, ANKH, LGALS3, PPP2R5B, GPX1, BNIP3L, BCL2L1, KLF7, FOSL2, ADM, MXI1, SELENBP1, PGF, FOXO3, NEDD4L, SLC2A1, SIAH2, MMP9, CXCR3, CD8A, IFNG, CD8B, JAK2, CCL4, ICAM1, WARSI, KRT1, GPR65, DYRK3, ACHE, ARG1, LCN2, LTF, OLFM4, HLA-DMB, YKT6, PDE4B, TWISTNB, BTN2A2,ZBTB33, PSMB9, CAMK4, TMEM19, SLC12A7, TP53BP1, PLEKHO1, SLC25A22, FRS2, CD24, STX1A, KCNMB4, CRISP2, HTRA1, PPL, RHBDF2, ZCCHC4, DDX6, SENP5, RAPGEF1, DTX2, RELB, DYRK2, CCNB1IP1, TDRD9, ZAP70, ARL14EP, MDC1, ADGRE3 and combinations thereof.

[0422] 201. The method of any one of Examples 198-200, wherein the composite biomarker score is selected from the group consisting of SeptiCyte LAB, Septicyte RAPID, Inflammatix TruVerity, and combinations thereof.

Claims

1. A method for generating sepsis risk indicators, the method comprising: Receive at least one input feature corresponding to the patient; The at least one input feature is analyzed using a machine learning model to generate a predictive score based on the at least one input feature, the predictive score indicating the probability that the patient has or develops sepsis within a predetermined time span; wherein the at least one input feature is associated with sepsis.

2. The method as described in claim 1, wherein, The at least one input feature includes clinical parameters.

3. The method as described in claim 1 or 2, wherein, The at least one input feature includes a biomarker.

4. The method according to any one of claims 1-3, further comprising: A diagnostic output is generated based on the predicted score.

5. The method according to any one of claims 1-4, wherein, The clinical parameters and / or the biomarkers are selected from the input feature groups listed in Table 8.

6. The method according to any one of claims 1-4, wherein, The clinical parameters and / or the biomarkers are selected from the input feature groups listed in Table 9.

7. The method according to any one of claims 1-4, wherein, The clinical parameters and / or the biomarkers are selected from the input feature groups listed in Table 10.

8. The method according to any one of claims 1-4, wherein, The clinical parameters and / or the biomarkers are selected from the input feature groups listed in Table 11.

9. The method according to any one of claims 1-4, wherein, The clinical parameters and / or the biomarkers are selected from the input feature groups listed in Table 12.

10. The method according to any one of claims 1-4, wherein, The clinical parameters and / or the biomarkers are selected from the input feature groups listed in Table 13.

11. The method according to any one of claims 5-10, wherein, The input feature groups listed in any of Tables 1-6 are listed according to their relative importance to the predicted score.

12. The method according to any one of claims 1-11, wherein, The relative importance of the predicted score is determined based on feature ranking, wherein the feature ranking is determined by ranking the input features based on the average SHAP value, wherein a high average SHAP value indicates a high relative importance to the predicted score, and a low SHAP value indicates a low relative importance to the predicted score.

13. The method of claim 12, wherein, The average SHAP value of the input features in any of the input feature groups listed in Table 8-13 is calculated using the following steps: Determine the absolute values ​​of all SHAP values ​​for a given population of the input features; Add the absolute values ​​of all SHAP values ​​to obtain the sum of the input features; as well as Divide the sum by the number of populations in the given population of the input features.

14. The method according to any one of claims 1-13, wherein, The at least one input feature includes procalcitonin, wherein the procalcitonin concentration values ​​are divided into two groups, wherein the relative importance of the first group of procalcitonin concentration values ​​is lower than that of the second group of procalcitonin concentration values.

15. The method of claim 14, wherein, The threshold for distinguishing between the first group of procalcitonin concentration values ​​and the second group of procalcitonin concentration values ​​is 2.5 log 10 pm / mL.

16. The method of claim 15, wherein, The relative importance of procalcitonin to the predicted score in each of the first and second groups is the median SHAP value of the procalcitonin concentration within the corresponding group.

17. The method according to any one of claims 1-13, wherein, The at least one input feature includes procalcitonin, wherein the procalcitonin concentration values ​​are divided into at least three groups, wherein the relative importance of the first group of procalcitonin concentration values ​​is lower than that of the second group of procalcitonin concentration values, and wherein the relative importance of the third group of procalcitonin concentration values ​​is higher than that of the first group of procalcitonin concentration values ​​and the second group of procalcitonin concentration values.

18. The method of claim 17, wherein, The first dividing threshold between the first group of procalcitonin concentration values ​​and the second group of procalcitonin concentration values ​​is 1.7 log10 pg / mL, and the second dividing threshold between the second group of procalcitonin concentration values ​​and the third group of procalcitonin concentration values ​​is 3.4 log10 pg / mL.

19. The method of claim 18, wherein, The relative importance of procalcitonin to the predicted score in each of the first, second, and third groups is the median SHAP value of the procalcitonin concentration within the corresponding group.

20. The method according to any one of claims 1-19, wherein, The input feature groups listed in Table 1 include: Glasgow Coma Scale score, FiO2, procalcitonin, systolic blood pressure, and platelet count.

21. The method of claim 20, wherein, The Glasgow Coma Scale score is more important than FiO2, FiO2 is more important than procalcitonin, procalcitonin is more important than systolic blood pressure, and systolic blood pressure is more important than platelet count.

22. The method according to any one of claims 1-19, wherein, The input feature groups listed in Table 1 include: Glasgow Coma Scale score, FiO2, procalcitonin, systolic blood pressure, platelet count, bilirubin, blood urea nitrogen, respiratory rate, diastolic blood pressure, and albumin.

23. The method according to any one of claims 1-19, wherein, The input feature groups listed in Table 1 include: Glasgow Coma Scale score, FiO2, procalcitonin, systolic blood pressure, platelet count, bilirubin, blood urea nitrogen, respiratory rate, diastolic blood pressure, albumin, creatinine, age, blood oxygen saturation, lactate, and C-reactive protein.

24. The method according to any one of claims 1-19, wherein, The input feature groups listed in Table 1 include: Glasgow Coma Scale score, FiO2, procalcitonin, systolic blood pressure, platelet count, bilirubin, blood urea nitrogen, respiratory rate, diastolic blood pressure, albumin, creatinine, age, blood oxygen saturation, lactate, C-reactive protein, potassium, neutrophil count, body temperature, total carbon dioxide, white blood cell count, lymphocyte count, sodium, heart rate, and chloride.

25. The method according to any one of claims 1-19, wherein, The input feature set includes procalcitonin and C-reactive protein.

26. The method of claim 25, wherein, One or both of procalcitonin and C-reactive protein are replaced by one or more other biomarkers listed in Table 2 based on their correlation values ​​associated with Table 3.

27. The method of claim 26, wherein, The correlation value of one or more other biomarkers that replace one or both of procalcitonin and C-reactive protein is less than XX.

28. The method of any one of claims 1-27, further comprising: The predicted score is determined to belong to one of four categories; as well as A diagnostic output is generated based on the category to which the predicted score belongs. The diagnostic output includes: the probability that the patient has or develops sepsis within the predetermined time span is low, moderate, high, or very high.

29. The method of claim 28, wherein, The predicted score ranges from 0 to 1.

30. The method of claim 28 or 29, wherein, The four categories include a low category, a medium category, a high category, and an extremely high category, wherein the low category is specified as the predicted score being below a first threshold, the medium category is specified as the predicted score being between the first threshold and a second threshold, the high category is specified as the predicted score being between the second threshold and a third threshold, and the extremely high category is specified as the predicted score being above the third threshold.

31. The method of claim 30, wherein, The predicted score is below the first threshold of 0.122, and the diagnostic output indicates that the probability of the patient having or developing sepsis within the predetermined time span is low.

32. The method of claim 30, wherein, The predicted score is equal to or higher than the first threshold of 0.122 and lower than the second threshold of 0.306, and the diagnostic output indicates that the patient has a moderate probability of having or developing sepsis within the predetermined time span.

33. The method of claim 30, wherein, The predicted score is equal to or higher than the second threshold of 0.306 and lower than the third threshold of 0.872, and the diagnostic output indicates that the patient has a high probability of having or developing sepsis within the predetermined time span.

34. The method of claim 30, wherein, The predicted score is equal to or higher than the third threshold of 0.872, and the diagnostic output indicates that the patient has an extremely high probability of having or developing sepsis within the predetermined time span.

35. The method according to any one of claims 1-34, further comprising: The predicted score is calibrated using a calibration model to generate a calibration score; as well as Risk categories are generated as diagnostic output based on the calibration score.

36. The method of claim 35, wherein, Generating the diagnostic output includes: The calibration score was determined to belong to one of four risk categories; and Based on the category to which the calibration score belongs, the risk category is generated, which includes: the probability that the patient has or develops sepsis within the predetermined time span is low, moderate, high, or very high.

37. The method of claim 36, wherein, The calibration score ranges from 0 to 1.

38. The method according to any one of claims 35-37, wherein, The risk category can be selected from a group consisting of low, medium, high and very high categories, wherein the low category is specified as the calibration score is below a first threshold, the medium category is specified as the calibration score is between the first and second thresholds, the high category is specified as the calibration score is between the second and third thresholds, and the very high category is specified as the calibration score is above the third threshold.

39. The method of claim 38, wherein, The calibration score is below the first threshold of 0.122, and the risk category indicates that the patient has a low probability of having or developing sepsis within the predetermined time span.

40. The method of claim 38, wherein, The calibration score is higher than the first threshold of 0.122 and lower than the second threshold of 0.306, and the risk category indicates that the patient has a moderate probability of having or developing sepsis within the predetermined time span.

41. The method of claim 38, wherein, The calibration score is higher than the second threshold of 0.306 and lower than the third threshold of 0.872, and the risk category indicates that the patient has a high probability of having or developing sepsis within the predetermined time span.

42. The method of claim 38, wherein, The calibration score is higher than the third threshold of 0.872, and the risk category indicates that the patient has an extremely high probability of having or developing sepsis within the predetermined time span.

43. The method according to any one of claims 1-42, wherein, The predetermined time span shall not exceed or exceed 12 hours, 18 hours, 24 hours, 36 hours, 48 ​​hours or 72 hours.

44. The method according to any one of claims 1-43, further comprising: At least one machine learning model is trained using training data, wherein the training data includes multiple biomarkers and health profiles of multiple patients and multiple subject diagnoses of the multiple patients.

45. The method according to any one of claims 4-44, wherein, The diagnostic output is a label automatically derived from the predicted score generated by the at least one machine learning model.

46. ​​The method according to any one of claims 4-45, wherein, The diagnostic output is determined by a physician based on the predicted score generated by the at least one machine learning model.

47. The method according to any one of claims 4-46, wherein, The analysis is performed without the value of one of the multiple clinical parameters and / or the value of one of the one or more biomarkers to generate the predictive score, and the method further includes generating the diagnostic output based on the predictive score.

48. The method of any one of claims 1-47, further comprising: Identify and impute the value of one of the input features, where the input feature is missing a value associated with the input feature.

49. The method according to any one of claims 1-47, wherein, One of the input features has a missing value, and the method further includes: The trigger logic decision module is used to determine whether to perform the analysis to generate the predicted score in the absence of the missing value of the input feature.

50. The method according to any one of claims 4-47, wherein, One of the input features has a missing value, and the method further includes: The missing values ​​of the input feature are imputed to generate imputed values ​​for the missing input features; Analyze data including the interpolated values ​​to generate the predictive score, which indicates the probability that the patient has or will develop sepsis within a predetermined time span; and The diagnostic output is generated based on the predicted score.

51. The method according to any one of claims 48-50, wherein, The imputation of missing values ​​is performed using a pre-trained template trained with training data, which includes input features, including multiple clinical parameters and multiple biomarkers of multiple patients, as well as multiple subject diagnoses of the multiple patients.

52. The method according to any one of claims 48-51, wherein, The input feature having the missing value is designated as the following feature: The value of this feature is measured or obtained outside a specified time window when data used to generate the predicted score is acceptable.

53. The method of claim 52, wherein, The specified time window is the same time period for the first group of input features, and is different for at least two input features in the second group of input features.

54. The method according to any one of claims 1-53, wherein, The input features include clinical parameters, which have values ​​measured within a specified time window.

55. The method according to any one of claims 1-54, wherein, The input features include biomarkers, which have measurements associated with the biomarker acquired within a specified time window.

56. The method according to any one of claims 1-55, wherein, The input features include clinical parameters, which include at least one of demographic measurements, patient assessments, vital signs, hematological laboratory values, and chemical laboratory values.

57. The method of any one of claims 48-50, further comprising: Identify and imput biomarkers that are missing values ​​associated with other biomarkers among the plurality of biomarkers.

58. The method according to any one of claims 48-57, wherein, The input feature that lacks a value associated with the input feature includes: biomarkers that lack a value associated with the biomarkers from the plurality of biomarkers, and / or clinical parameters that lack a value associated with the clinical parameters from the plurality of clinical parameters.

59. The method according to any one of claims 1-58, wherein, The input features include at least 20 clinical parameters and at least two biomarkers.

60. The method according to any one of claims 5-10, wherein, The input feature groups listed in any of Tables 1-6 include at least one biomarker feature and at least one clinical parameter feature, wherein the values ​​of one or more biomarker features are divided into two groups, and the values ​​of one or more clinical parameter features are divided into two groups.

61. The method of claim 60, wherein, The biomarker features include at least one of procalcitonin, C-reactive protein, interleukin-6, pentaprotein 3, interleukin-8, interleukin-1 receptor antagonist protein, and vascular cell adhesion molecule 1, wherein the relative importance of the first set of concentration values ​​of each biomarker feature is lower than that of the second set of concentration values ​​of the same biomarker feature.

62. The method of claim 60 or 61, wherein, The clinical parameters include at least one of platelets, systolic blood pressure, diastolic blood pressure, albumin, and blood oxygen saturation, wherein the relative importance of the first set of values ​​for each clinical parameter feature is higher than the relative importance of the second set of concentration values ​​for that clinical parameter feature.

63. The method of claim 60 or 61, wherein, The clinical parameters include at least one of respiratory rate, blood urea nitrogen, bilirubin and creatinine, and wherein the relative importance of a first set of values ​​for each clinical parameter feature is lower than that of a second set of concentration values ​​for the same clinical parameter feature.

64. The method of claim 60, wherein, The biomarker features include procalcitonin, wherein the relative importance of the first group of procalcitonin concentration values ​​is lower than that of the second group of procalcitonin concentration values.

65. The method of claim 60, wherein, The biomarker features include C-reactive protein, wherein the relative importance of the concentration values ​​of the first group of C-reactive protein is lower than that of the concentration values ​​of the second group of C-reactive protein.

66. The method of claim 60, wherein, The biomarker features include interleukin-6, wherein the relative importance of the first group of interleukin-6 concentration values ​​is lower than that of the second group of interleukin-6 concentration values.

67. The method of claim 60, wherein, The biomarker features include pentaprotein 3, wherein the relative importance of the first group of pentaprotein 3 concentration values ​​is lower than that of the second group of pentaprotein 3 concentration values.

68. The method of claim 60, wherein, The biomarker features include interleukin-8, wherein the relative importance of the first group of interleukin-8 concentration values ​​is lower than that of the second group of interleukin-8 concentration values.

69. The method of claim 60, wherein, The biomarker features include interleukin-1 receptor antagonist proteins, wherein the relative importance of the concentration values ​​of the first group of interleukin-1 receptor antagonist proteins is lower than that of the concentration values ​​of the second group of interleukin-1 receptor antagonist proteins.

70. The method of claim 60, wherein, The biomarker features include vascular cell adhesion molecule 1, wherein the relative importance of the concentration values ​​of the first group of vascular cell adhesion molecule 1 is lower than that of the concentration values ​​of the second group of vascular cell adhesion molecule 1.

71. The method of claim 60, wherein, The biomarker features include procalcitonin and C-reactive protein, wherein the relative importance of the first group of procalcitonin concentration values ​​is lower than that of the second group of procalcitonin concentration values, and the relative importance of the first group of C-reactive protein concentration values ​​is lower than that of the second group of C-reactive protein concentration values.

72. The method of claim 60, wherein, The biomarker features include procalcitonin and interleukin-6, wherein the relative importance of the first group of procalcitonin concentration values ​​is lower than that of the second group of procalcitonin concentration values, and the relative importance of the first group of interleukin-6 concentration values ​​is lower than that of the second group of interleukin-6 concentration values.

73. The method of claim 60, wherein, The biomarker features include interleukin-6 and C-reactive protein, wherein the relative importance of the first group of interleukin-6 concentration values ​​is lower than that of the second group of interleukin-6 concentration values, and the relative importance of the first group of C-reactive protein concentration values ​​is lower than that of the second group of C-reactive protein concentration values.

74. The method of claim 60, wherein, The biomarker features include procalcitonin, C-reactive protein, and interleukin-6, wherein the relative importance of the procalcitonin concentration values ​​in the first group is lower than that in the second group, the relative importance of the C-reactive protein concentration values ​​in the first group is lower than that in the second group, and the relative importance of the interleukin-6 concentration values ​​in the first group is lower than that in the second group.

75. The method of claim 60, wherein, The biomarker features include pentaprotein 3 and interleukin-1 receptor antagonist protein, wherein the relative importance of the concentration values ​​of the first group of pentaprotein 3 is lower than that of the second group of pentaprotein 3, and the relative importance of the concentration values ​​of the first group of interleukin-1 receptor antagonist protein is lower than that of the second group of interleukin-1 receptor antagonist protein.

76. The method of claim 60, wherein, The biomarker features include pentaprotein 3 and interleukin 8, wherein the relative importance of the first group of pentaprotein 3 concentration values ​​is lower than that of the second group of pentaprotein 3 concentration values, and the relative importance of the first group of interleukin 8 concentration values ​​is lower than that of the second group of interleukin 8 concentration values.

77. The method of claim 60, wherein, The biomarker features include procalcitonin and interleukin-8, wherein the relative importance of the first group of procalcitonin concentration values ​​is lower than that of the second group of procalcitonin concentration values, and the relative importance of the first group of interleukin-8 concentration values ​​is lower than that of the second group of interleukin-8 concentration values.

78. The method of claim 60, wherein, The biomarker features include procalcitonin and pentaprotein 3, wherein the relative importance of the first group of procalcitonin concentration values ​​is lower than that of the second group of procalcitonin concentration values, and the relative importance of the first group of pentaprotein 3 concentration values ​​is lower than that of the second group of pentaprotein 3 concentration values.

79. The method of claim 60, wherein, The biomarker features include pentaprotein 3 and interleukin 6, wherein the relative importance of the first group of pentaprotein 3 concentration values ​​is lower than that of the second group of pentaprotein 3 concentration values, and the relative importance of the first group of interleukin 6 concentration values ​​is lower than that of the second group of interleukin 6 concentration values.

80. The method of claim 60, wherein, The clinical parameter features include platelets, with the relative importance of the first group of platelet values ​​being higher than that of the second group of platelet values.

81. The method of claim 60, wherein, The clinical parameter features include systolic blood pressure, with the first group of systolic blood pressure values ​​having a higher relative importance than the second group of systolic blood pressure values.

82. The method of claim 60, wherein, The clinical parameter features include diastolic blood pressure, with the first group of diastolic blood pressure values ​​having a higher relative importance than the second group of diastolic blood pressure values.

83. The method of claim 60, wherein, The clinical parameter features include blood oxygen saturation, with the relative importance of the first group of blood oxygen saturation values ​​being higher than that of the second group of blood oxygen saturation values.

84. The method of claim 60, wherein, The clinical parameter features include albumin, with the relative importance of the first group of albumin values ​​being higher than that of the second group of albumin values.

85. The method of claim 60, wherein, The clinical parameter features include respiratory rate, wherein the relative importance of the first group of respiratory rate values ​​is lower than that of the second group of respiratory rate values.

86. The method of claim 60, wherein, The clinical parameter features include blood urea nitrogen, with the relative importance of the first group of blood urea nitrogen values ​​being lower than that of the second group of blood urea nitrogen values.

87. The method of claim 60, wherein, The clinical parameter features include bilirubin, with the relative importance of the first group of bilirubin values ​​being lower than that of the second group of bilirubin values.

88. The method of claim 60, wherein, The clinical parameter features include creatinine, with the relative importance of the first group of creatinine values ​​being lower than that of the second group of creatinine values.

89. The method of claim 60, wherein, The clinical parameters include creatinine and blood urea nitrogen, wherein the relative importance of creatinine values ​​in the first group is lower than that in the second group, and the relative importance of blood urea nitrogen values ​​in the first group is lower than that in the second group.

90. The method of claim 60, wherein, The clinical parameters include platelets, creatinine, and blood urea nitrogen, wherein the relative importance of the first group of platelet values ​​is higher than that of the second group of platelet values, the relative importance of the first group of creatinine values ​​is lower than that of the second group of creatinine values, and the relative importance of the first group of blood urea nitrogen values ​​is lower than that of the second group of blood urea nitrogen values.

91. The method of claim 60, wherein, The clinical parameters include respiratory rate, creatinine, and blood urea nitrogen, wherein the relative importance of the first group of respiratory rate values ​​is lower than that of the second group of respiratory rate values, the relative importance of the first group of creatinine values ​​is lower than that of the second group of creatinine values, and the relative importance of the first group of blood urea nitrogen values ​​is lower than that of the second group of blood urea nitrogen values.

92. The method of claim 60, wherein, The clinical parameters include platelet count and respiratory rate, wherein the relative importance of the first group of platelet count values ​​is higher than that of the second group of platelet count values, and the relative importance of the first group of respiratory rate values ​​is lower than that of the second group of respiratory rate values.

93. The method of claim 60, wherein, The clinical parameters include platelet count and systolic blood pressure, wherein the relative importance of the first group of platelet count values ​​is higher than that of the second group of platelet count values, and the relative importance of the first group of systolic blood pressure values ​​is lower than that of the second group of systolic blood pressure values.

94. The method of claim 60, wherein, The clinical parameters include platelet count, respiratory rate, and systolic blood pressure, wherein the relative importance of the first group of platelet count values ​​is higher than that of the second group of platelet count values, the relative importance of the first group of respiratory rate values ​​is lower than that of the second group of respiratory rate values, and the relative importance of the first group of systolic blood pressure values ​​is higher than that of the second group of systolic blood pressure values.

95. The method of claim 60, wherein, The relative importance of one or more features in each of the first and second groups to the predicted score is the median SHAP value of the feature values ​​within the corresponding group.

96. The method of claim 95, wherein, The relative importance of one or more biomarker features in each of the first and second groups to the predicted score is the median SHAP value of the biomarker feature values ​​within the corresponding group.

97. The method of claim 95, wherein, The relative importance of one or more clinical parameter features to the predicted score in each of the first and second groups is the median SHAP value of the clinical parameter feature values ​​within the corresponding group.

98. The method of claim 60, wherein, The relative importance of one or more features in each of the first and second groups to the predicted score is the average SHAP value of the feature values ​​within the corresponding group.

99. The method of claim 98, wherein, The relative importance of one or more biomarker features in each of the first and second groups to the predicted score is the average SHAP value of the biomarker feature values ​​within the corresponding group.

100. The method of claim 98, wherein, The relative importance of one or more clinical parameter features to the predicted score in each of the first and second groups is the average SHAP value of the clinical parameter feature values ​​within the corresponding group.

101. The method according to any one of claims 61-100, wherein, The threshold for distinguishing between the first group of procalcitonin concentration values ​​and the second group of procalcitonin concentration values ​​is 2.24 log 10 pg / mL.

102. The method according to any one of claims 61-100, wherein, The threshold for dividing the pentameric protein 3 concentration values ​​in the first group and the second group is 3.98 log10 pg / mL.

103. The method according to any one of claims 61-100, wherein, The threshold for distinguishing between the interleukin-8 concentration values ​​in the first group and the interleukin-8 concentration values ​​in the second group is 1.41 log10 pg / mL.

104. The method according to any one of claims 61-100, wherein, The threshold for distinguishing between the concentration values ​​of interleukin-1 receptor antagonists in the first group and the concentration values ​​of interleukin-1 receptor antagonists in the second group is 3.48 log10 pg / mL.

105. The method according to any one of claims 61-100, wherein, The threshold for distinguishing between the interleukin-6 concentration values ​​in the first group and the interleukin-6 concentration values ​​in the second group is 2.07 log10 pg / mL.

106. The method according to any one of claims 61-100, wherein, The threshold for dividing the C-reactive protein concentration values ​​of the first group and the second group is 7.63 log10 pg / mL.

107. The method according to any one of claims 61-100, wherein, The threshold for distinguishing between the concentration values ​​of vascular cell adhesion molecule 1 in the first group and the concentration values ​​of vascular cell adhesion molecule 1 in the second group is 6.17 log10 pg / mL.

108. The method according to any one of claims 61-100, wherein, The threshold for distinguishing between the first group of platelet concentration values ​​and the second group of platelet concentration values ​​is 162 × 10⁹ / L.

109. The method according to any one of claims 61-100, wherein, The threshold for distinguishing between the first group of systolic blood pressure values ​​and the second group of systolic blood pressure values ​​is 101 mmHg.

110. The method according to any one of claims 61-100, wherein, The threshold for dividing the first group of respiratory rate values ​​into the second group of respiratory rate values ​​is 101 breaths per minute.

111. The method according to any one of claims 61-100, wherein, The threshold for distinguishing between the blood urea nitrogen (BUN) values ​​in the first group and the BUN values ​​in the second group is 24 mg / dL.

112. The method according to any one of claims 61-100, wherein, The threshold for distinguishing between the first group of bilirubin values ​​and the second group of bilirubin values ​​is 1.19 mg / dL.

113. The method according to any one of claims 61-100, wherein, The threshold for dividing the creatinine values ​​in the first group and the second group is 1.56 mg / dL.

114. The method according to any one of claims 61-100, wherein, The threshold for distinguishing between the first and second diastolic blood pressure values ​​is 54.5 mmHg.

115. The method according to any one of claims 61-100, wherein, The threshold for distinguishing between the first group of blood oxygen saturation values ​​and the second group of blood oxygen saturation values ​​is 91%.

116. The method according to any one of claims 5-10, wherein, The input feature groups listed in any of Tables 1-6 include at least one biomarker feature and at least one clinical parameter feature, wherein the values ​​of one or more biomarker features are divided into at least three groups, and the values ​​of one or more clinical parameter features are divided into at least three groups.

117. The method of claim 116, wherein, The biomarker features include at least one of procalcitonin, C-reactive protein, interleukin-6, pentamin-3, interleukin-8, interleukin-1 receptor antagonist protein, and vascular cell adhesion molecule 1, wherein the relative importance of the first set of concentration values ​​of each biomarker feature is lower than that of the second set of concentration values ​​of the biomarker feature, and wherein the relative importance of the third set of concentration values ​​of the biomarker feature is higher than that of the first set of concentration values ​​and the second set of concentration values.

118. The method of claim 116, wherein, The clinical parameters include at least one of platelets, systolic blood pressure, diastolic blood pressure, albumin, and pulse oxygen saturation, wherein the relative importance of a first set of values ​​for each clinical parameter feature is higher than that of a second set of concentration values ​​for the same clinical parameter feature, and wherein the relative importance of a third set of concentration values ​​for the same clinical parameter feature is lower than that of the first set of clinical parameter values ​​and the second set of clinical parameter values.

119. The method of claim 118, wherein, The clinical parameters include at least one of respiratory rate, blood urea nitrogen, bilirubin and creatinine, wherein the relative importance of a first set of values ​​for each clinical parameter feature is lower than that of a second set of concentration values ​​for the same clinical parameter feature, and wherein the relative importance of a third set of values ​​for each clinical parameter feature is higher than that of the first set of clinical parameter feature values ​​and the second set of clinical parameter feature values.

120. The method of claim 116, wherein, The biomarker features include procalcitonin, wherein the relative importance of the first group of procalcitonin concentration values ​​is lower than that of the second group of procalcitonin concentration values, and wherein the relative importance of the third group of procalcitonin concentration values ​​is higher than that of the first group of procalcitonin concentration values ​​and the second group of procalcitonin concentration values.

121. The method of claim 116, wherein, The biomarker features include C-reactive protein, wherein the relative importance of the concentration values ​​of the first group of C-reactive protein is lower than that of the second group of C-reactive protein, and wherein the relative importance of the concentration values ​​of the third group of C-reactive protein is higher than that of the first group of C-reactive protein and the second group of C-reactive protein.

122. The method of claim 116, wherein, The biomarker features include interleukin-6, wherein the relative importance of the first group of interleukin-6 concentration values ​​is lower than that of the second group of interleukin-6 concentration values, and wherein the relative importance of the third group of interleukin-6 concentration values ​​is higher than that of the first group of interleukin-6 concentration values ​​and the second group of interleukin-6 concentration values.

123. The method of claim 116, wherein, The biomarker features include pentamericin 3, wherein the relative importance of the first group of pentamericin 3 concentration values ​​is lower than that of the second group of pentamericin 3 concentration values, and wherein the relative importance of the third group of pentamericin 3 concentration values ​​is higher than that of the first group of pentamericin 3 concentration values ​​and the second group of pentamericin 3 concentration values.

124. The method of claim 116, wherein, The biomarker features include interleukin-8, wherein the relative importance of the first group of interleukin-8 concentration values ​​is lower than that of the second group of interleukin-8 concentration values, and wherein the relative importance of the third group of interleukin-8 concentration values ​​is higher than that of the first group of interleukin-8 concentration values ​​and the second group of interleukin-8 concentration values.

125. The method of claim 116, wherein, The biomarker features include interleukin-1 receptor antagonist proteins, wherein the relative importance of the concentration values ​​of the first group of interleukin-1 receptor antagonist proteins is lower than that of the second group of interleukin-1 receptor antagonist proteins, and wherein the relative importance of the concentration values ​​of the third group of interleukin-1 receptor antagonist proteins is higher than that of the first group of interleukin-1 receptor antagonist proteins and the second group of interleukin-1 receptor antagonist proteins.

126. The method of claim 116, wherein, The biomarker features include vascular cell adhesion molecules 1, wherein the relative importance of the concentration values ​​of the first group of vascular cell adhesion molecules 1 is lower than that of the second group of vascular cell adhesion molecules 1, and wherein the relative importance of the concentration values ​​of the third group of vascular cell adhesion molecules 1 is higher than that of the first group of vascular cell adhesion molecules 1 and the second group of vascular cell adhesion molecules 1.

127. The method of claim 116, wherein, The biomarker features include procalcitonin and C-reactive protein, wherein the relative importance of the first group of procalcitonin concentration values ​​is lower than that of the second group of procalcitonin concentration values, and the relative importance of the third group of procalcitonin concentration values ​​is higher than that of the first and second groups of procalcitonin concentration values, and the relative importance of the first group of C-reactive protein concentration values ​​is lower than that of the second group of C-reactive protein concentration values, and the relative importance of the third group of C-reactive protein concentration values ​​is higher than that of the first and second groups of C-reactive protein concentration values.

128. The method of claim 116, wherein, The biomarker characteristics include procalcitonin and interleukin-6, wherein the relative importance of the first group of procalcitonin concentration values ​​is lower than that of the second group of procalcitonin concentration values, and the relative importance of the third group of procalcitonin concentration values ​​is higher than that of the first group of procalcitonin concentration values ​​and the second group of procalcitonin concentration values.

129. The method of claim 116, wherein, The biomarker features include C-reactive protein and interleukin-6, wherein the relative importance of the concentration values ​​of the first group of C-reactive protein is lower than that of the second group of C-reactive protein, and the relative importance of the concentration values ​​of the third group of C-reactive protein is higher than that of the first group of C-reactive protein and the second group of C-reactive protein. Similarly, the relative importance of the concentration values ​​of the first group of interleukin-6 is lower than that of the second group of interleukin-6, and the relative importance of the concentration values ​​of the third group of interleukin-6 is higher than that of the first group of interleukin-6 and the second group of interleukin-6.

130. The method of claim 116, wherein, The biomarker characteristics include procalcitonin, C-reactive protein, and interleukin-6, wherein the relative importance of the first group of procalcitonin concentration values ​​is lower than that of the second group of procalcitonin concentration values, and the relative importance of the third group of procalcitonin concentration values ​​is higher than that of the first and second groups of procalcitonin concentration values. The relative importance of the first group of C-reactive protein concentration values ​​is lower than that of the second group of C-reactive protein concentration values, and the relative importance of the third group of C-reactive protein concentration values ​​is higher than that of the first and second groups of C-reactive protein concentration values. The relative importance of the first group of interleukin-6 concentration values ​​is lower than that of the second group of interleukin-6 concentration values, and the relative importance of the third group of interleukin-6 concentration values ​​is higher than that of the first and second groups of interleukin-6 concentration values.

131. The method of claim 116, wherein, The biomarker features include pentaprotein 3 and interleukin-1 receptor antagonist protein, wherein the relative importance of the concentration values ​​of the first group of pentaprotein 3 is lower than that of the second group of pentaprotein 3, and the relative importance of the concentration values ​​of the third group of pentaprotein 3 is higher than that of the first group of pentaprotein 3 and the second group of pentaprotein 3.

132. The method of claim 116, wherein, The biomarker features include pentaprotein 3 and interleukin 8, wherein the relative importance of the first group of pentaprotein 3 concentration values ​​is lower than that of the second group of pentaprotein 3 concentration values, and the relative importance of the third group of pentaprotein 3 concentration values ​​is higher than that of the first group of pentaprotein 3 concentration values ​​and the second group of pentaprotein 3 concentration values.

133. The method of claim 116, wherein, The biomarker characteristics include procalcitonin and interleukin-8, wherein the relative importance of the first group of procalcitonin concentration values ​​is lower than that of the second group of procalcitonin concentration values, and the relative importance of the third group of procalcitonin concentration values ​​is higher than that of the first group of procalcitonin concentration values ​​and the second group of procalcitonin concentration values. The relative importance of the first group of interleukin-8 concentration values ​​is lower than that of the second group of interleukin-8 concentration values, and the relative importance of the third group of interleukin-8 concentration values ​​is higher than that of the first group of interleukin-8 concentration values ​​and the second group of interleukin-8 concentration values.

134. The method of claim 116, wherein, The biomarker features include procalcitonin and pentaprotein 3, wherein the relative importance of the first group of procalcitonin concentration values ​​is lower than that of the second group of procalcitonin concentration values, and the relative importance of the third group of procalcitonin concentration values ​​is higher than that of the first group of procalcitonin concentration values ​​and the second group of procalcitonin concentration values.

135. The method of claim 116, wherein, The biomarker features include pentaprotein 3 and interleukin 6, wherein the relative importance of the first group of pentaprotein 3 concentration values ​​is lower than that of the second group of pentaprotein 3 concentration values, and the relative importance of the third group of pentaprotein 3 concentration values ​​is higher than that of the first group of pentaprotein 3 concentration values ​​and the second group of pentaprotein 3 concentration values.

136. The method of claim 116, wherein, The clinical parameter features include platelets, wherein the relative importance of the first group of platelet values ​​is higher than that of the second group of platelet values, and wherein the relative importance of the third group of platelet values ​​is lower than that of the first group of platelet values ​​and the second group of platelet values.

137. The method of claim 116, wherein, The clinical parameter features include systolic blood pressure, wherein the relative importance of the first group of systolic blood pressure values ​​is higher than that of the second group of systolic blood pressure values, and wherein the relative importance of the third group of systolic blood pressure values ​​is lower than that of the first group of systolic blood pressure values ​​and the second group of systolic blood pressure values.

138. The method of claim 116, wherein, The clinical parameter features include diastolic blood pressure, wherein the relative importance of the first group of diastolic blood pressure values ​​is higher than that of the second group of diastolic blood pressure values, and wherein the relative importance of the third group of diastolic blood pressure values ​​is lower than that of the first group of diastolic blood pressure values ​​and the second group of diastolic blood pressure values.

139. The method of claim 116, wherein, The clinical parameter features include pulse oxygen saturation, wherein the relative importance of the first group of pulse oxygen saturation values ​​is higher than that of the second group of pulse oxygen saturation values, and the relative importance of the third group of pulse oxygen saturation values ​​is lower than that of the first group of pulse oxygen saturation values ​​and the second group of pulse oxygen saturation values.

140. The method of claim 116, wherein, The clinical parameter features include albumin, wherein the relative importance of the first group of albumin values ​​is higher than that of the second group of albumin values, and wherein the relative importance of the third group of albumin values ​​is lower than that of the first group of albumin values ​​and the second group of albumin values.

141. The method of claim 116, wherein, The clinical parameter features include respiratory rate, wherein the relative importance of the first group of respiratory rate values ​​is higher than that of the second group of respiratory rate values, and wherein the relative importance of the third group of respiratory rate values ​​is lower than that of the first group of respiratory rate values ​​and the second group of respiratory rate values.

142. The method of claim 116, wherein, The clinical parameter features include blood urea nitrogen, wherein the relative importance of blood urea nitrogen values ​​in the first group is higher than that in the second group, and the relative importance of blood urea nitrogen values ​​in the third group is lower than that in the first and second groups.

143. The method of claim 116, wherein, The clinical parameter features include bilirubin, wherein the relative importance of the first group of bilirubin values ​​is higher than that of the second group of bilirubin values, and wherein the relative importance of the third group of bilirubin values ​​is lower than that of the first group of bilirubin values ​​and the second group of bilirubin values.

144. The method of claim 116, wherein, The clinical parameter features include creatinine, wherein the relative importance of creatinine values ​​in the first group is higher than that in the second group, and the relative importance of creatinine values ​​in the third group is lower than that in the first and second groups.

145. The method of claim 116, wherein, The clinical parameters include creatinine and blood urea nitrogen, wherein the relative importance of the first group of creatinine values ​​is higher than that of the second group of creatinine values, and the relative importance of the third group of creatinine values ​​is lower than that of the first and second groups of creatinine values.

146. The method of claim 116, wherein, The clinical parameters include platelets, creatinine, and blood urea nitrogen. The relative importance of the first group of platelet values ​​is higher than that of the second group, and the relative importance of the third group of platelet values ​​is lower than that of the first and second groups. The relative importance of the first group of creatinine values ​​is higher than that of the second group, and the relative importance of the third group of creatinine values ​​is lower than that of the first and second groups. The relative importance of the first group of blood urea nitrogen values ​​is higher than that of the second group, and the relative importance of the third group of blood urea nitrogen values ​​is lower than that of the first and second groups.

147. The method of claim 116, wherein, The clinical parameters include respiratory rate, creatinine, and blood urea nitrogen. The relative importance of the first group of respiratory rate values ​​is higher than that of the second group, and the relative importance of the third group of respiratory rate values ​​is lower than that of the first and second groups. The relative importance of the first group of creatinine values ​​is higher than that of the second group, and the relative importance of the third group of creatinine values ​​is lower than that of the first and second groups. The relative importance of the first group of blood urea nitrogen values ​​is higher than that of the second group, and the relative importance of the third group of blood urea nitrogen values ​​is lower than that of the first and second groups.

148. The method of claim 116, wherein, The clinical parameters include platelet count and respiratory rate, wherein the relative importance of the first group of platelet count values ​​is higher than that of the second group of platelet count values, and the relative importance of the third group of platelet count values ​​is lower than that of the first group of platelet count values ​​and the second group of platelet count values. The relative importance of the first group of respiratory rate values ​​is higher than that of the second group of respiratory rate values, and the relative importance of the third group of respiratory rate values ​​is lower than that of the first group of respiratory rate values ​​and the second group of respiratory rate values.

149. The method of claim 116, wherein, The clinical parameters include platelet count and systolic blood pressure, wherein the relative importance of the first group of platelet count values ​​is higher than that of the second group of platelet count values, and the relative importance of the third group of platelet count values ​​is lower than that of the first group of platelet count values ​​and the second group of platelet count values. The relative importance of the first group of systolic blood pressure values ​​is higher than that of the second group of systolic blood pressure values, and the relative importance of the third group of systolic blood pressure values ​​is lower than that of the first group of systolic blood pressure values ​​and the second group of systolic blood pressure values.

150. The method of claim 116, wherein, The clinical parameters include platelets, respiratory rate, and systolic blood pressure. The relative importance of the first group of platelet values ​​is higher than that of the second group, and the relative importance of the third group of platelet values ​​is lower than that of the first and second groups. The relative importance of the first group of respiratory rate values ​​is higher than that of the second group, and the relative importance of the third group of respiratory rate values ​​is lower than that of the first and second groups. The relative importance of the first group of systolic blood pressure values ​​is higher than that of the second group, and the relative importance of the third group of systolic blood pressure values ​​is lower than that of the first and second groups.

151. The method of claim 116, wherein, The relative importance of one or more features in each of the first, second, and third groups to the predicted score is the median SHAP value of the feature values ​​within the corresponding group.

152. The method of claim 151, wherein, The relative importance of one or more biomarker features in each of the first, second, and third groups to the predicted score is the median SHAP value of the biomarker feature values ​​within the corresponding group.

153. The method of claim 151, wherein, The relative importance of one or more clinical parameter features to the predicted score in each of the first, second, and third groups is the median SHAP value of the clinical parameter feature values ​​within the corresponding group.

154. The method of claim 116, wherein, The relative importance of one or more features in each of the first, second, and third groups to the predicted score is the average SHAP value of the feature values ​​within the corresponding group.

155. The method of claim 154, wherein, The relative importance of one or more biomarker features in each of the first, second, and third groups to the predicted score is the average SHAP value of the biomarker feature values ​​within the corresponding group.

156. The method of claim 154, wherein, The relative importance of one or more clinical parameter features to the predicted score in each of the first, second, and third groups is the average SHAP value of the clinical parameter feature values ​​within the corresponding group.

157. The method according to any one of claims 116-156, wherein, The threshold for dividing the procalcitonin concentration values ​​in the first group and the second group is 2.05 log10 pg / mL, and the second threshold for dividing the procalcitonin concentration values ​​in the second group and the third group is 2.85 log10 pg / mL.

158. The method according to any one of claims 116-156, wherein, The threshold for dividing the pentameric protein 3 concentration values ​​in the first group and the second group is 3.61 log 10 pg / mL, and the second threshold for dividing the pentameric protein 3 concentration values ​​in the second group and the third group is 3.98 log 10 pg / mL.

159. The method according to any one of claims 116-156, wherein, The threshold for dividing the interleukin-8 concentration values ​​of the first group and the second group is 1.36 log10 pg / mL, and the second threshold for dividing the interleukin-8 concentration values ​​of the second group and the third group is 1.97 log10 pg / mL.

160. The method according to any one of claims 116-156, wherein, The threshold for dividing the concentration values ​​of interleukin-1 receptor antagonists in the first group and the second group is 3.15 log10 pg / mL, and the second threshold for dividing the concentration values ​​of interleukin-1 receptor antagonists in the second group and the third group is 3.57 log10 pg / mL.

161. The method according to any one of claims 116-156, wherein, The threshold for dividing the interleukin-6 concentration values ​​of the first group and the second group is 3.15 log10 pg / mL, and the second threshold for dividing the interleukin-6 concentration values ​​of the second group and the third group is 3.57 log10 pg / mL.

162. The method according to any one of claims 116-156, wherein, The threshold for dividing the C-reactive protein concentration values ​​of the first group and the second group is 7.31 log10 pg / mL, and the second threshold for dividing the C-reactive protein concentration values ​​of the second group and the third group is 7.91 log10 pg / mL.

163. The method according to any one of claims 116-156, wherein, The threshold for dividing the concentration values ​​of the first group of vascular cell adhesion molecules 1 and the second group of vascular cell adhesion molecules 1 is 6.05 log 10 pg / mL, and the second threshold for dividing the concentration values ​​of the second group of vascular cell adhesion molecules 1 and the third group of vascular cell adhesion molecules 1 is 6.17 log 10 pg / mL.

164. The method according to any one of claims 116-156, wherein, The threshold for dividing the platelet concentration values ​​of the first group and the second group is 139 10^9 / L, and the second threshold for dividing the platelet concentration values ​​of the second group and the third group is 188 10^9 / L.

165. The method according to any one of claims 116-156, wherein, The threshold for dividing the first group of systolic blood pressure values ​​into the second group of systolic blood pressure values ​​is 91 mmHg, and the second threshold for dividing the second group of systolic blood pressure values ​​into the third group of systolic blood pressure values ​​is 111 mmHg.

166. The method according to any one of claims 116-156, wherein, The threshold for dividing the first group of respiratory rate values ​​into the second group of respiratory rate values ​​is 23.5 breaths per minute, and the second threshold for dividing the second group of respiratory rate values ​​into the third group of respiratory rate values ​​is 28 breaths per minute.

167. The method according to any one of claims 116-156, wherein, The threshold for dividing the blood urea nitrogen (BUN) values ​​between the first group and the second group is 21 mg / dL, and the second threshold for dividing the blood urea nitrogen values ​​between the second group and the third group is 41 mg / dL.

168. The method according to any one of claims 116-156, wherein, The threshold for dividing the first group of bilirubin values ​​into the second group of bilirubin values ​​is 0.88 mg / dL, and the second threshold for dividing the second group of bilirubin values ​​into the third group of bilirubin values ​​is 1.5 mg / dL.

169. The method according to any one of claims 116-156, wherein, The threshold for dividing the creatinine values ​​in the first group and the second group is 1.15 mg / dL, and the second threshold for dividing the creatinine values ​​in the second group and the third group is 1.73 mg / dL.

170. The method according to any one of claims 116-156, wherein, The threshold for dividing the first group of diastolic blood pressure values ​​into the second group of diastolic blood pressure values ​​is 52 mmHg, and the second threshold for dividing the second group of diastolic blood pressure values ​​into the third group of diastolic blood pressure values ​​is 66 mmHg.

171. The method according to any one of claims 116-156, wherein, The threshold for dividing the first group of blood oxygen saturation values ​​into the second group of blood oxygen saturation values ​​is 81 mmHg, and the second threshold for dividing the second group of blood oxygen saturation values ​​into the third group of blood oxygen saturation values ​​is 92.5 mmHg.

172. A method wherein the method of any one of claims 4-171 generates the diagnostic output based on the predicted score, classifying the patient as having sepsis or at risk of developing sepsis within a predetermined time span, the method further comprising: The individual was given a therapeutically effective dose of sepsis treatment.

173. The method of claim 172, wherein, Based on the generated diagnostic output, the treatment includes at least one of antibiotics or intravenous infusion.

174. A method for treating an individual with sepsis, the individual having a score based on a plurality of input features, the score indicating the probability that the patient has or will develop sepsis based on the method of any one of claims 1-171, the treatment comprising administering an effective amount of one or more sepsis therapies to the individual.

175. A method comprising measuring a plurality of input features relating to an individual and / or samples from an individual in Tables 8, 9, 10, 11, 12 or 13, optionally wherein the measurements of at least a first input feature and at least a second input feature are sequential, and the measurement of the second input feature depends on the measurement of the first input feature.

176. The method of claim 175, comprising: The majority of the input features in Table 8 concerning the individual.

177. The method of claim 175 or 176, wherein, The individual is suspected of having sepsis, is suspected of developing sepsis, or is at risk of developing sepsis, or the individual has one or more symptoms of sepsis.

178. The method according to any one of claims 175-177, wherein, The individual was suspected of developing sepsis within 24 hours of being measured.

179. The method according to any one of claims 175-178, wherein, The samples are blood, urine, tissue, skin, saliva, sputum, and / or mucus.

180. The method according to any one of claims 175-179, wherein, The individual in question has an infection.

181. The method of claim 180, wherein, The infection is a viral infection, a bacterial infection, a fungal infection, or a prion infection.

182. The method of claim 180 or 181, wherein, The infection is SARS-CoV-2, influenza, meningitis, pneumonia, tuberculosis, or hepatitis.

183. The method according to any one of claims 175-182, wherein, The individuals are older than approximately 65 years and younger than approximately 1 year, have weakened immune function, suffer from chronic diseases, have recently been seriously ill or hospitalized, and / or have a history of sepsis.

184. The method of claim 183, wherein, The chronic diseases mentioned are diabetes, lung disease, cancer, heart disease, hepatitis, or kidney disease.

185. A method comprising measuring 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, or 13 of any one or more of Tables 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, or more of input features about an individual.

186. The method of claim 185, wherein, The individual has one or more symptoms of sepsis.

187. The method of claim 185 or 186, wherein, The individual is suspected of having sepsis, is suspected of developing sepsis, or is at risk of developing sepsis.

188. The method according to any one of claims 185-187, wherein, The individual was suspected of developing sepsis within 24 hours of a measurement taken from the individual.

189. The method according to any one of claims 185-188, wherein, The individual in question has an infection.

190. The method of claim 189, wherein, The infection is a viral infection, bacterial infection, fungal infection, or prion infection.

191. The method of claim 189 or 190, wherein, The infection is SARS-CoV-2, influenza, meningitis, pneumonia, tuberculosis, or hepatitis.

192. The method according to any one of claims 185-191, wherein, The individuals are older than approximately 65 years and younger than approximately 1 year, have weakened immune function, suffer from chronic diseases, have recently been seriously ill or hospitalized, and / or have a history of sepsis.

193. The method of claim 192, wherein, The chronic diseases mentioned are diabetes, lung disease, cancer, heart disease, or kidney disease.

194. The method according to any one of claims 1-171, wherein, The multiple input features include at least one input feature from the first group, at least one input feature from the second group, and at least one input feature from the third group.

195. The method of claim 194, in, The first group includes input features selected from a set of the following: age, sex at birth, race, ethnicity, patient history, current chief complaint or symptoms, neurological assessment, clinical decision support alerts, clinician or other chart records, diagnostic codes, surgeries performed on the patient, current medications, interventions, patient care environment, medical imaging data or assessments, electrocardiogram, endoscopy, systolic blood pressure, diastolic blood pressure, body temperature, respiratory rate, heart rate, blood oxygen saturation, inhaled oxygen concentration, and combinations thereof; The second group includes input features selected from a set of the following: white blood cell count, lymphocyte count, neutrophil count, platelet count, creatinine, blood urea nitrogen, potassium, chloride, total carbon dioxide, sodium, albumin, bilirubin, lactate, glucose, calcium, hematocrit, hemoglobin concentration, laboratory-derived biomarkers, clinical severity indicators and composite indicators, biopsy data, microbial test data and combinations thereof; and The third group includes input features selected from a set of the following: procalcitonin, C-reactive protein, protein biomarkers, genomic biomarkers, gene expression or transcriptome biomarkers, composite biomarker scores and combinations thereof.

196. The method of claim 194 or 195, wherein, The first group includes input features selected from a set consisting of: Vital signs (e.g., body temperature, heart rate, blood pressure, respiratory rate, blood oxygen saturation); Demographic information (e.g., age, sex, race, ethnicity); Past medical history; Current chief complaint or symptoms; Neurological assessment (Glasgow Coma Scale, Four-Nutrient Assessment (FOUR)); Clinical decision support alerts; Clinician or other chart / graph records; Diagnostic codes (e.g., ICD-10); Surgery (including CPT codes); Current medication (including NDC code); Intervention measures; Patient care environment (ICU, ED, wards, outpatient department, etc.); Imaging examinations (e.g., X-ray, CT scan, MRI, fMRI, ultrasound); Electrocardiogram (e.g., EEG, EKG); Endoscopic examination; and The combination of the above items.

197. The method according to any one of claims 194-196, wherein, The second group includes input features selected from a set consisting of the following: Laboratory tests; Laboratory-derived biomarkers (e.g., monocyte distribution width, Cytovale IntelliSep); Clinical severity indicators and comprehensive indicators (e.g., Sequential Organ Failure Assessment (SOFA) score, SOFA component scores, Charlson Comorbidity Index, Acute Physiology and Chronic Health Assessment (APACHE) I and II); Biopsy; Microbial testing (culture, PCR testing, antigen testing); and The combination of the above items.

198. The method according to any one of claims 194-196, wherein, The third group includes input features selected from a set of the following: Protein biomarkers; Genomic biomarkers; Gene expression or transcriptome biomarkers; Composite biomarker scoring; and The combination of the above items.

199. The method of claim 198, wherein, The protein biomarkers are selected from the group consisting of: angiopoietin-1, angiopoietin-2, C-reactive protein, cystatin C, D-dimer, E-selectin, chemokine, FLT3 ligand, GCSF, GDF15, GMCSF, granzyme B, IFN-α, IFN-γ, IL1-β, IL1-RA, IL-2, IL-4, IL-6, IL-7, IL-8, IL-10, IL-15, IP-10, lactate dehydrogenase (LDH), lipopolysaccharide-binding protein (LBP), leptin, MCP1, MIP1-α, MIP1-β, MIP3-α, NGAL, islet calculiin, PDL-1, pentaprotein-3, procalcitonin, protein C, Sl00b, TGF-α, thrombomodulin, tissue factor, TNF-α, TRAIL, TREM-1, troponin, VCAM-1, VEG-F, and combinations thereof.

200. The method of claim 198 or 199, wherein, The gene expression or transcriptome biomarkers are selected from the group consisting of the following: CEACAM4, LAMP1, PLAC8, PLA2G7, CEACAM1, ZDHHC19, C9orf95, GNA15, BATF, C3AR1, KIAA1370, TGFBI, MTCH1, RPGRIP1, HLA-DPB1, nuclear factor-κB (NF-κB), heme oxygenase-1 (HO-1), FCGR2C, CRP, IL-6, IL-10, IL-12, TNF, LY6E, LAMP3, ISG15, USP18, RSAD2, IFI44L, NCOA7, CMPK2, IFI44, BA TF2, GBP4, IFIH1, GMPR, CXCL10, IFI27, ISG15, CCL2, OAS3, P2RY14, GCH1, CD274, ID01, ARL4A, BPGM, TNFAIP3, TNFAIP6, IL1R1, PROK2, RGS1, CXCR6, RTP 4. TLR3, FZD5, KCNJ2, IRAK2, RGS16, CXCL8, OLR1, CCRL2, SPP1, ST3GAL4, CST7, IL1R2, IL4R, GALM, CTLA4, SOCS1, BCL2L1, TGM2, SLC39A8, CA2, PNP, GATA 1. SLC1A5, RGS16, CD44, PDGFC, CSF2RA, CXCL1, CSF3R, SOCS3, IL18R1, PIM1, LEPR, IL1B, CSF2RB, ITGB3, STAT1, FAS, EFNA1, IER3, EETS2, SLC2A3, BCL2 A1, PFKFB3, CEBPB, BIRC2, ATF3, TUBB2A, G0S2, MERTK, BIK, QSOX1, PYGL, TGFA, P4HA2, LDHA, IRS2, TSPO, HGF, GADD45A, TGFBR3, CD38, PRF1, FASLG, TIMP 3. ANKH, LGALS3, PPP2R5B, GPX1, BNIP3L, BCL2L1, KLF7, FOSL2, ADM, MXI1, SELENBP1, PGF, FOXO3, NEDD4L, SLC2A1, SIAH2, MMP9, CXCR3, CD8A, IFNG, CD8B , JAK2, CCL4, ICAM1, WARSI, KRT1, GPR65, DYRK3, ACHE, ARG1, LCN2, LTF, OLFM4, HLA-DMB, YKT6, PDE4B, TWISTNB, BTN2A2, ZBTB33, PSMB9, CAMK4, TMEM19,SLC12A7, TP53BP1, PLEKHO1, SLC25A22, FRS2, CD24, STX1A, KCNMB4, CRISP2, HTRA1, PPL, RHBDF2, ZCCHC4, DDX6, SENP5, RAPGEF1, DTX2, RELB, DYRK2, CCNB1IP1, TDRD9, ZAP70, ARL14EP, MDC1, ADGRE3 and combinations thereof.

201. The method according to any one of claims 198-200, wherein, The composite biomarker score is selected from the group consisting of the following: SeptiCyte LAB, Septicyte RAPID, Inflammatix TruVerity, and combinations thereof.

Images