Septemia typing division method, typing prediction model construction method and device, and typing prediction model application method and device

Abstract

The invention discloses a sepsis typing division method and a typing prediction model construction and application method and device, and relates to the field of machine learning, and the method comprises the steps: obtaining central venous oxygen saturation and peripheral perfusion indexes of a plurality of patients in a preset time period; calculating and acquiring a dynamic blood flow oxygen flow index track of each patient in the preset time period according to the central venous oxygen saturation and the peripheral perfusion index of the patient in the preset time period; evaluating trajectory similarity by using a dynamic time warping algorithm, and grouping trajectories by using a hierarchical clustering algorithm to obtain a plurality of sepsis types; and training the machine learning model by using each sepsis type and the corresponding clinical features of the patient to obtain a sepsis type prediction model. According to the application, sepsis typing division is carried out based on the longitudinal data of the central venous oxygen saturation and the peripheral perfusion index, a prediction model is established for each typing based on clinical features, and finally division and accurate prediction of sepsis crowds are realized.

Description

Technical Field

[0001] The present application relates to the field of machine learning, and in particular to a method for classifying sepsis, building a classification prediction model, and an application method and device. Background Art

[0002] Sepsis is defined as life-threatening organ dysfunction caused by the body's dysregulated response to infection. Sepsis is often accompanied by systemic inflammation, dysregulated immune responses, and metabolic disturbances. In severe cases, it can rapidly deteriorate into septic shock, causing widespread organ failure and a high mortality rate. Despite significant advances in the definition, diagnostic criteria, and treatment of sepsis in recent years, it remains one of the most lethal acute illnesses worldwide. The mortality rate from sepsis varies among different populations but is generally estimated to be between 30% and 50%.

[0003] Because sepsis is a highly heterogeneous syndrome, identifying distinct clinical subtypes is crucial for precise treatment and management. Early studies on sepsis subtype classification relied primarily on clinical features. However, these clinically based classifications are often based on cross-sectional data (at a single point in time) and fail to accurately reflect the dynamic evolution of sepsis. In recent years, researchers have begun exploring the classification of sepsis subtypes based on longitudinal data that can reflect dynamic clinical evolution. For example, a sepsis subtype classification characterized by dynamic body temperature trajectories has been proposed. Temperature patterns may reflect distinct inflammatory responses, thereby influencing the clinical presentation and mortality of sepsis. Furthermore, the trajectory of the Sequential Organ Failure Assessment (SOFA) score has also been used to analyze sepsis subtypes. Longitudinal data studies are beneficial for revealing the characteristics of sepsis, a disease that evolves rapidly over a short period of time. However, the development of a classification method with good predictive power and clinical feasibility using longitudinal data remains an ongoing research project. Summary of the Invention

[0004] The purpose of this application is to provide a sepsis classification method, classification prediction model construction, application method and device, which can calculate and obtain the dynamic blood flow oxygen flow index based on the dynamic central venous oxygen saturation and peripheral perfusion index, perform sepsis classification, and then use each sepsis classification and its corresponding patient clinical characteristics to train a machine learning model to achieve accurate prediction of sepsis classification.

[0005] To achieve the above objectives, this application provides the following solutions: In a first aspect, the present application provides a method for classifying sepsis, comprising: Obtain central venous oxygen saturation and peripheral perfusion index of several patients within a preset time period; Obtaining a trajectory of the dynamic blood flow oxygen index of each patient within a preset time period based on the patient's central venous oxygen saturation and peripheral perfusion index within a preset time period; the blood flow oxygen index is calculated based on the central venous oxygen saturation and peripheral perfusion index; The dynamic time warping algorithm was used to evaluate the similarity of the blood flow oxygen flow index trajectories of several patients; According to the similarity of the blood flow oxygen flow index trajectories, the hierarchical clustering algorithm was used to group the dynamic blood flow oxygen flow index trajectories and obtain multiple sepsis types.

[0006] In a second aspect, the present application provides a method for constructing a sepsis typing prediction model, comprising: A machine learning model is trained using each sepsis classification and its corresponding patient clinical characteristics to derive a sepsis classification prediction model; the sepsis classification prediction model is used to predict the sepsis classification based on the patient's clinical characteristics; each sepsis classification is determined based on the above-mentioned sepsis classification method.

[0007] In a third aspect, the present application provides a method for applying a sepsis classification prediction model, comprising: Obtain the patient's clinical characteristics within the target time period; The clinical characteristics of the patient within the target time period are input into the sepsis classification prediction model to obtain the sepsis classification; the sepsis classification prediction model is obtained based on the above-mentioned method for constructing a sepsis classification prediction model based on dynamic blood flow and oxygen flow.

[0008] In a fourth aspect, the present application provides a computer device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-mentioned sepsis classification method, or the above-mentioned sepsis classification prediction model construction method, or the above-mentioned sepsis classification prediction model application method.

[0009] In a fifth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned sepsis classification method, or the above-mentioned sepsis classification prediction model construction method, or the above-mentioned sepsis classification prediction model application method.

[0010] According to the specific embodiments provided in this application, this application discloses the following technical effects: The present application provides a method for sepsis classification, classification prediction model construction, application method and device. The present application calculates a dynamic blood flow oxygen flow index based on central venous oxygen saturation and peripheral perfusion index, and then uses a dynamic time warping algorithm and a hierarchical clustering algorithm to cluster and group the trajectory of the dynamic blood flow oxygen flow index to obtain multiple sepsis classifications, thereby discovering / proving that the clinical outcomes of patients of each classification are different. Based on this, a machine learning model is trained to obtain a sepsis classification prediction model. The present application classifies sepsis based on longitudinal data of central venous oxygen saturation and peripheral perfusion index, and establishes a prediction model for the classification based on clinical characteristics, ultimately achieving the classification and accurate prediction of the sepsis population. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0012] Figure 1 This is a diagram of an application environment of a sepsis classification method in one embodiment of the present application; Figure 2 A schematic diagram of a flow chart of a method for classifying sepsis types provided in one embodiment of the present application; Figure 3 A schematic diagram of a method for screening patients required for sepsis classification according to an embodiment of the present application; Figure 4 A schematic diagram of a 72-hour dynamic blood flow oxygen flow index trajectory curve for a training set provided in one embodiment of the present application; Figure 5 A schematic diagram of a 72-hour dynamic blood flow oxygen flow index trajectory curve for a validation set provided in one embodiment of the present application; Figure 6 A histogram of the proportions of each group in the training set and the validation set provided in an embodiment of the present application; Figure 7 A schematic diagram of the PISO index trajectory of a training set when clustering using the Time Warping Algorithm (HAC) + Hierarchical Clustering (DTW) method provided in one embodiment of the present application; Figure 8 A schematic diagram of the PISO index trajectory of a training set when clustering using the group-based trajectory modeling (GBTM) method provided in one embodiment of the present application; Figure 9 A schematic diagram of sensitivity analysis of the HAC+DTW method and the GBTM method for classification provided in one embodiment of the present application; Figure 10A SHAP (SHapley Additive exPlanations) diagram corresponding to type 1 provided in an embodiment of the present application; Figure 11 The SHAP value corresponding to type 1 provided in an embodiment of the present application; Figure 12 This is the SHAP diagram corresponding to type 2 provided in one embodiment of the present application; Figure 13 The SHAP value corresponding to type 2 provided in an embodiment of the present application; Figure 14 This is the SHAP diagram corresponding to type 3 provided in one embodiment of the present application; Figure 15 The SHAP value corresponding to type 3 provided in an embodiment of the present application; Figure 16 A schematic diagram of a process for applying a sepsis classification prediction model provided in one embodiment of the present application; Figure 17 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0013] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0014] Central venous oxygen saturation (ScvO2) and peripheral perfusion index (PI) are two critical hemodynamic parameters that can be obtained in real time. Central venous oxygen saturation, measured by blood gas analysis of superior vena cava blood, reflects the balance between oxygen delivery and utilization; peripheral perfusion index, obtained by a blood monitor, reflects peripheral tissue perfusion. These two indices have been widely used in clinical hemodynamic assessment, and numerous studies have demonstrated their respective value in critically ill patients and their ability to predict outcomes in patients with sepsis. However, few studies have investigated sepsis classification based on hemodynamic characteristics, and no study has combined central venous oxygen saturation and peripheral perfusion index to classify sepsis. Furthermore, many studies on sepsis classification are based on cross-sectional data, which cannot reflect the rapidly evolving characteristics of sepsis over a short period of time.

[0015] In this regard, the present application proposes a sepsis classification method, classification prediction model construction, application method and device, which classifies sepsis based on the dynamic blood flow oxygen flow index calculated from the central venous oxygen saturation and peripheral perfusion index, and then uses each sepsis classification and its corresponding patient clinical characteristics to train a machine learning model to achieve accurate prediction of sepsis subtypes.

[0016] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0017] The sepsis classification method provided in the embodiment of the present application is specifically a sepsis classification method based on dynamic blood flow and oxygen flow, which can be applied to Figure 1 In the application environment shown. Among them, the terminal communicates with the server through the network. The data storage system can store the data that the server needs to process. The data storage system can be set up separately, integrated on the server, or placed on the cloud or other servers. The terminal can send the central venous oxygen saturation and peripheral perfusion index of several patients within a preset time period to the server. After the server receives the central venous oxygen saturation and peripheral perfusion index of several patients within the preset time period, the server calculates the dynamic blood flow oxygen flow index trajectory of each patient within the preset time period based on the central venous oxygen saturation and peripheral perfusion index of the patient within the preset time period; the blood flow oxygen flow index is calculated based on the central venous oxygen saturation and peripheral perfusion index. The dynamic time warping algorithm is used to evaluate the similarity of the blood flow oxygen flow index trajectories of several patients. Based on the similarity of the blood flow oxygen flow index trajectories, the hierarchical clustering algorithm is used to group the blood flow oxygen flow index trajectories to obtain multiple sepsis classifications. In addition, in some embodiments, the sepsis classification method can also be implemented independently by a server or a terminal. For example, the terminal can directly determine the sepsis classification, or the server can obtain the patient's central venous oxygen saturation and peripheral perfusion index from a data storage system and perform sepsis classification.

[0018] The terminals may be, but are not limited to, various desktop computers, laptops, smart phones, tablet computers, IoT devices, and portable wearable devices. The server may be implemented as an independent server or a server cluster consisting of multiple servers, or as a cloud server.

[0019] In an exemplary embodiment, Figure 2 As shown, a method for classifying sepsis is provided. The method is executed by a computer device, specifically a computer device such as a terminal or a server, or a terminal and a server. In the embodiment of the present application, the method is applied to Figure 1The server in is used as an example to illustrate, including the following steps 101 to 105.

[0020] Step 101: Obtain central venous oxygen saturation and peripheral perfusion index of several patients within a preset time period.

[0021] Step 102, calculate and obtain the trajectory of the dynamic blood flow oxygen index of each patient within the preset time period based on the patient's central venous oxygen saturation and peripheral perfusion index within the preset time period; the blood flow oxygen index is calculated based on the central venous oxygen saturation and peripheral perfusion index.

[0022] Step 103 : Calculate the trajectory similarity of the blood flow oxygen flow index of several patients using a dynamic time warping algorithm.

[0023] Step 104 : Based on the similarity of the blood flow oxygen flow index trajectories, a hierarchical clustering algorithm is used to group the blood flow oxygen flow index trajectories to obtain multiple sepsis classifications.

[0024] By implementing steps 101 to 104 above, there are currently few studies on sepsis classification based on hemodynamic characteristics, and no study has yet combined central venous oxygen saturation and peripheral perfusion index to classify sepsis. In this application, a dynamic blood flow oxygen index is calculated based on central venous oxygen saturation and peripheral perfusion index, and the dynamic blood flow oxygen index trajectory is clustered using a dynamic time warping algorithm and a clustering algorithm to derive multiple sepsis classifications. It is found / proven that the clinical outcomes of patients in each classification are different, and various types of sepsis can be accurately classified.

[0025] Data collected in Step 101 were obtained from the Intensive Care Unit (ICU) of Peking Union Medical College Hospital, including structured medical records such as the medical record cover, admission notes, and medical history. Patients were admitted between January 1, 2015, and December 31, 2023. Inclusion criteria included patients aged 18 years or older. Exclusion criteria included missing key demographic information (such as age or sex) or the absence of 12 consecutive measurements of central venous oxygen saturation and peripheral perfusion index (measured every 6 hours) within 72 hours of sepsis diagnosis.

[0026] The total number of patients who met the inclusion and exclusion criteria for this application was 12,975, 10,525 of whom developed defined infections after admission to the ICU, and 5,205 of whom met the criteria for organ failure in the time window before and after the onset of infection. Finally, 720 patients were able to provide central venous oxygen saturation and peripheral perfusion index measurement data at 12 consecutive time points (once every 6 hours) within 72 hours after the diagnosis of sepsis. The specific enrollment process is as follows: Figure 3shown.

[0027] Define diagnostic criteria for sepsis and screen patients for sepsis based on these criteria so that data from these patients can be used to categorize sepsis. According to the Sepsis 3.0 standard, sepsis is defined as infection and associated organ failure. The specific definition and screening process is as follows: (1) Occurrence of infection: The patient develops an infection after admission to the ICU and meets the following conditions within a specific time window: first, a pathogen culture order is issued (regardless of whether the culture result is negative or positive); second, the patient receives therapeutic antibiotic treatment. Therapeutic antibiotics are defined as the continuous use of a single antibiotic for more than 48 hours. Regarding the specific time window, if the pathogen culture order is issued before the therapeutic antibiotic, the therapeutic antibiotic must be started within 72 hours after the pathogen culture order is issued; if the pathogen culture order is issued after the therapeutic antibiotic, the pathogen culture must be performed within 24 hours after the start of the antibiotic, and the time when the therapeutic antibiotic is started is used as the "onset of infection time."

[0028] (2) Organ failure: Within a certain time window before and after the defined onset of infection (48 hours before and 24 hours after the onset of infection), the patient's Sequential Organ Dysfunction Assessment (SOFA) score increases by at least 2 points compared to the baseline value before admission to the ICU. The SOFA score (Sequential Organ Dysfunction Assessment) is a scale used to assess the degree of organ failure in patients with sepsis, covering subscores of six major organ systems: respiratory, coagulation, liver, cardiovascular, central nervous system, and kidney. Each subscore is calculated every 6 hours based on the patient's data within the first 72 hours after the patient is admitted to the ICU, and the worst variable value during this period is used to assess the score. When calculating the SOFA score, special attention should be paid to the following: for the assessment of the central nervous system, the lowest Glasgow Coma Scale (GCS) is used, which is assessed by the patient's state before sedation. The baseline value of the SOFA score is based on the patient's SOFA score before admission to the ICU; if the patient did not undergo relevant biochemical tests before admission to the ICU, his or her SOFA baseline value is uniformly considered to be 0.

[0029] In step 102, based on the connotations of central venous oxygen saturation and peripheral perfusion index, the present application innovatively designs the PISO index (blood flow oxygen index) and its formula: ,in, Represents the critical value based on PI to predict the patient's adverse prognosis (death, mechanical ventilation, ICU stay, etc.). This formula allows the simultaneous comparison of PI and ScvO2, and the prediction of adverse outcomes based on PI and critical value. The direction of the value is confirmed.

[0030] To establish The value can make the PISO index effectively predict the patient's poor prognosis. According to the range of critical values ​​of central venous oxygen saturation and peripheral perfusion index reported in existing literature, it is assumed that 0.7, 0.8, 0.9, 1.0 and 1.1 respectively, and then explored different The PISO index trajectories of the population were clustered into three types using the hierarchical clustering (HAC) method and the dynamic time warping (DTW) method. The variance analysis was used to compare the differences in the 28-day mortality of patients in each type. By calculating the F value and P value, the influence of each parameter value on the prognostic index after clustering was obtained. When the value is 1.1, the three types can have the greatest prognostic difference, thus determining The optimal value is 1.1, as shown in Table 1. In ANOVA, the F-value and P-value are two core statistics used to determine whether differences between groups are significant. The F-value measures the extent to which the mean differences between groups are relative to the individual variability within the groups. The P-value indicates the probability of observing the current or more extreme F-value, assuming the null hypothesis is true.

[0031] Table 1 is different Differences in various prognostic indicators under different values In Table 1 , for the data corresponding to the duration of mechanical ventilation and ICU hospitalization days within 28 days, the values ​​before the brackets [] are the medians, and the values ​​in the brackets [] are the first quartile and the third quartile, respectively.

[0032] In steps 103 and 104, after the PISO index calculation formula was established, data for each enrolled patient consisted of 12 PISO index vectors generated every 6 hours over a 72-hour period. The study population was randomly divided into a training set and a validation set in a 4:1 ratio to internally validate the model. The similarity of PISO index trajectories in the training set was then assessed using the DTW method, and the HAC method was used to group PISO index trajectories based on similarity.

[0033] In order to determine the optimal number of groups for classification, this application used a variety of clustering performance evaluation indicators, including silhouette coefficient (Silhouette Coefficient), db index (Davies-Bouldin index), variance ratio (Variance Ratio), BIC (Bayesian Information Criterion), WCSS (Within-Cluster Sum of Squares), Hartigan's test (Hartigan's test) and SD index (Standard Deviation index) for comparison, and finally determined that the optimal number of groups was 3, as shown in Table 2, i.e., 3 types of classification.

[0034] Table 2 Clustering performance evaluation

[0035] Subsequently, the HAC+DTW method was repeated in the validation set, and it was found that the grouping results were consistent between the training set and the validation set. Figures 4 to 6 As shown, Figure 4 shows the number of cluster groups of the training set, Figure 5 shows the number of cluster groups for the validation set, Figure 6 A histogram of the number of people in each cluster group between the two queues is shown.

[0036] In addition, to ensure the robustness of the derived classification, this application further adopts the group-based trajectory modeling method (GBTM method). The GBTM method is a latent class analysis method that calculates the probability of each patient belonging to different classifications through maximum likelihood estimation. The sensitivity analysis results of the HAC+DTW method and the GBTM method show that the two methods have good consistency, such as Figures 7 to 9 shown.

[0037] Observing the trajectory characteristics of each phenotype, in the training set, phenotype 1 had the fewest patients (n=140, 24.3%), characterized by a persistent increase in the PISO index within the first 54 hours after the diagnosis of sepsis, followed by a decrease. Genotype 2 (n=246, 42.7%) had a trajectory characterized by a slow increase in the PISO index from the onset of sepsis to within 72 hours, making it the most common phenotype. Genotype 3 (n=190, 33.0%) was characterized by a slow decrease in the PISO index from a negative baseline value within 72 hours. Similar trajectory trends were observed for each of these phenotypes in the validation set. Here, n represents the number of patients in the corresponding phenotype, and the percentage represents the percentage of patients in that phenotype relative to the total number of patients.

[0038] In another exemplary embodiment of the present application, a method for constructing a sepsis classification prediction model is provided, comprising: A machine learning model is trained using each sepsis classification and its corresponding patient clinical characteristics to generate a sepsis classification prediction model. The sepsis classification prediction model is used to predict the sepsis classification based on the patient's clinical characteristics. Each sepsis classification is determined based on the sepsis classification method described in steps 101 to 104 above.

[0039] In this application, multiple sepsis classifications were analyzed and then used as a basis for training a machine learning model to derive a sepsis classification prediction model. This application uses the blood flow oxygen flow index calculated from central venous oxygen saturation and peripheral perfusion index to classify sepsis. This can draw on the real-time, longitudinal data, and easy availability of central venous oxygen saturation and peripheral perfusion index to accurately predict sepsis classification.

[0040] Based on the three PISO index-based classifications established in the previous steps, the differences in prognosis after diagnosis of sepsis among patients of each classification were subsequently compared. Prognostic indicators included the incidence of acute kidney injury (AKI), ICU length of stay, mechanical ventilation duration, and 28-day mortality. The results showed that patients with classification 3 had the worst prognosis, with significant differences in AKI incidence, ICU length of stay, mechanical ventilation duration, and 28-day mortality compared with patients of the other two classifications (P < 0.05). As shown in Table 3, although patients with classification 1 showed a trend of being better than patients with classification 2 in all of the above indicators, no difference in prognosis was found between patients with classification 1 and patients with classification 2. Therefore, this application discovered and demonstrated the clinical significance of sepsis classification.

[0041] Table 3 Differences in prognosis among patients with different types of sepsis

[0042] The three sepsis classifications showed significant differences in clinical presentation within 6 hours of sepsis diagnosis (baseline), as shown in Table 4. Based on these differences, multiple machine learning models were constructed and trained to predict a patient's blood flow and oxygen flow classification using clinical features at the onset of sepsis. A dataset consisting of clinical features and corresponding classifications of several sepsis patients comprised 80% of the training set, with the remaining 20% ​​used as a test set. The test sample sizes for classifications 1, 2, and 3 were 35, 62, and 47, respectively. Model performance was evaluated using precision, recall, and F1 score, and the results are summarized in Table 5. The K-nearest neighbor (KNN) model performed the worst, exhibiting low recognition ability across all test samples. The support vector machine (SVM) performed slightly better than the KNN, but overall still lagged behind more complex models. The multilayer perceptron (MLP) model improved on both the KNN and SVM models, demonstrating superior classification capabilities. Among the various tree ensemble-based methods, the performance differences among XGBoost, Random Forest, and LightGBM are relatively small, but Random Forest performs the best among the three models. Specifically, the F1 scores of Random Forest in the three types were 0.78, 0.82, and 0.93, respectively, showing high accuracy and stability. Especially in the identification of type 3, the F1 value was as high as 0.93, showing extremely strong predictive ability. Based on the classification results of the random forest model, the SHAP graph was further used to show the direction and importance of the influence of the top 20 features on the model output when predicting the three types (type 1, type 2, and type 3) based on the PISO index. Figures 10 to 15 As shown in the SHAP diagram, blue to red represents the size of the eigenvalue, and PI shows a distribution trend from smaller eigenvalues ​​(blue) to larger eigenvalues ​​(red and purple) from the left to the right of the coordinate axis, indicating that high-value PI features have a positive impact on the model prediction results. Figure 10 、 Figure 12 and Figure 14 The “SHAP value” in represents the impact on the model output; Figure 11 、 Figure 13 and Figure 15 "mean ( )” represents the average impact on the model output size.

[0043] Table 4 Statistical analysis results of clinical characteristics In Table 4, mean (SD) represents the mean (standard deviation); median [Q1, Q3] represents the median (first quartile and third quartile). The units corresponding to the clinical characteristics in Table 4 are: age (years), ScvO2 (%), body mass index (BMI) (kg / m 2 ), body temperature (℃), heart rate (beats / min), respiratory rate (beats / min), mean arterial pressure (mmHg), systolic blood pressure (mmHg), diastolic blood pressure (mmHg), lymphocyte percentage (%), mean corpuscular hemoglobin (g / L), prothrombin time (s), indirect bilirubin ( ), albumin concentration (g / L), urea nitrogen concentration (12.3mmol / L), platelet concentration (10 9 / L), white blood cell count (10 9 / L), hematocrit (%), fibrinogen (g / L), platelet distribution width (fl), mean platelet volume (fl), lymphocyte count (10 9 / L), activated partial thromboplastin time (seconds), absolute neutrophil count (10 9 / L), platelet volume (%), activated partial thromboplastin time ratio (%), D-dimer concentration (mg / L), creatinine concentration ( ), alanine aminotransferase (U / L), positive end-expiratory pressure (PEEP) (cmH2O), direct bilirubin ( ), total bilirubin ( ), procalcitonin (ng / m l ).

[0044] Table 5 Performance results of each model in the test set

[0045] Therefore, in step 205, each sepsis classification and its corresponding patient clinical characteristics are used to train a machine learning model to obtain a sepsis classification prediction model, which specifically includes: (1) Use each sepsis classification and its corresponding patient clinical characteristics to train multiple machine learning models.

[0046] (2) Evaluate the performance of each machine learning model after training.

[0047] (3) The trained machine learning model with the best performance was selected as the sepsis classification prediction model.

[0048] Furthermore, the method for constructing a sepsis classification prediction model further includes: (1) Based on the classification results of the best-performing trained machine learning model, the SHAP graph was combined to determine the direction and importance of the impact of each clinical feature on the model output.

[0049] (2) Determine the key clinical features for predicting sepsis classification based on the direction and importance of each clinical feature's impact on the model output. Key clinical features refer to the clinical features ranked highest in importance in the SHAP graph corresponding to each clinical feature.

[0050] Figure 7 All clinical features in the model are used as model input to establish the model and analyze the key clinical features that have a greater impact on the prediction results. When predicting in actual scenarios, the final model input selection can be based on the clinical features ranked first in importance in the SHAP graph for prediction, and the specific number can be defined manually.

[0051] In this application, sepsis subtypes are classified based on central venous oxygen saturation and peripheral perfusion index, which are clinically available in real time. Specifically, the PISO index is calculated using central venous oxygen saturation and peripheral perfusion index. Similarity and clustering calculations are then performed on the PISO index trajectory, ultimately resulting in three subtypes. This results in a sepsis classification prediction model that can predict sepsis subtypes based on the PISO index. This application provides a reliable sepsis classification method and sepsis classification prediction model based on readily available, real-time, and dynamic clinical data.

[0052] In another exemplary embodiment of the present application, Figure 16 As shown, a method for applying a sepsis classification prediction model is provided, comprising: S1: Obtain the patient's clinical characteristics within a target time period. The clinical characteristics within the target time period are those ranked by importance in the SHAP plot corresponding to each clinical characteristic. As an example, clinical characteristics within 6 hours of sepsis diagnosis or other time periods may be obtained.

[0053] S2: Inputting the clinical characteristics of the patient within the target time period into the sepsis classification prediction model to obtain the sepsis classification; the sepsis classification prediction model is obtained based on the above-mentioned sepsis classification prediction model construction method.

[0054] The present application also provides an application scenario, which applies the above-mentioned sepsis classification method, sepsis classification prediction model construction method and the above-mentioned sepsis classification prediction model application method. Specifically, it can be applied in the patient sepsis classification prediction scenario. This scenario includes a sepsis classification classification link, a sepsis classification prediction model construction link and a patient sepsis classification prediction link. The sepsis classification construction link is used to calculate the PISO index based on the central venous oxygen saturation and peripheral perfusion index of the septic patient, and after clustering, obtain a variety of sepsis classifications based on the PISO index; the sepsis classification prediction model construction link is used to train a machine learning model based on this classification to construct a sepsis classification prediction model; the patient sepsis classification prediction link is used to apply the sepsis classification prediction model according to the clinical characteristics of the septic patient to obtain the corresponding sepsis classification.

[0055] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 17 As shown. The computer device includes a processor, memory, an input / output (I / O) interface, and a communication interface. The processor, memory, and I / O interface are connected via a system bus, and the communication interface is connected to the system bus via the I / O interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store relevant data required for constructing sepsis typing and a sepsis typing prediction model, as well as the ultimately constructed sepsis typing prediction model. The I / O interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal via a network connection. When executed by the processor, the computer program implements the aforementioned sepsis typing method, the aforementioned sepsis typing prediction model construction method, or the aforementioned sepsis typing prediction model application method.

[0056] Those skilled in the art will understand that Figure 17The structure shown in the figure is merely a block diagram of a portion of the structure related to the present application solution and does not constitute a limitation on the computer device to which the present application solution is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement. In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, it implements the above-mentioned sepsis classification method, or the above-mentioned sepsis classification prediction model construction method, or the above-mentioned sepsis classification prediction model application method.

[0057] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, which, when executed by a processor, implements the above-mentioned sepsis classification method, or the above-mentioned sepsis classification prediction model construction method, or the above-mentioned sepsis classification prediction model application method.

[0058] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0059] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0060] This document uses specific examples to illustrate the principles and implementation methods of this application. The description of the above examples is only intended to help understand the method and core concept of this application. At the same time, for those skilled in the art, based on the concept of this application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for classifying sepsis, characterized in that: include: Obtain central venous oxygen saturation and peripheral perfusion index of several patients within a preset time period; Calculate and obtain the trajectory of the dynamic blood flow oxygen index of each patient within a preset time period based on the patient's central venous oxygen saturation and peripheral perfusion index within a preset time period; the blood flow oxygen index is calculated based on the central venous oxygen saturation and peripheral perfusion index; The dynamic time warping algorithm was used to evaluate the similarity of the blood flow oxygen flow index trajectories of several patients; According to the similarity of the blood flow oxygen flow index trajectories, the hierarchical clustering algorithm was used to group the blood flow oxygen flow index trajectories and obtain multiple sepsis types.

2. The method for classifying sepsis according to claim 1, wherein: The calculation formula of the blood flow oxygen flow index is: ； Wherein, PISO represents blood flow oxygen index; PI represents peripheral perfusion index; ScvO2 represents central venous oxygen saturation; Represents the critical value for predicting poor prognosis of patients based on peripheral perfusion index PI.

3. A method for constructing a sepsis typing prediction model, characterized in that: include: A machine learning model is trained using each sepsis classification and its corresponding patient clinical characteristics to obtain a sepsis classification prediction model; the sepsis classification prediction model is used to predict the sepsis classification based on the patient's clinical characteristics; each sepsis classification is determined based on the sepsis classification method described in any one of claims 1-2.

4. The method for constructing a sepsis typing prediction model according to claim 3, wherein: The machine learning model was trained using each sepsis classification and its corresponding patient clinical characteristics to derive a sepsis classification prediction model, specifically including: Using each sepsis classification and its corresponding patient clinical characteristics to train multiple machine learning models; Evaluate the performance of each trained machine learning model; The trained machine learning model with the best performance was selected as the sepsis classification prediction model.

5. The method for constructing a sepsis typing prediction model according to claim 4, wherein: The method for constructing a sepsis typing prediction model further includes: Based on the classification results of the best-performing trained machine learning model, the SHAP graph was combined to determine the direction and importance of each clinical feature's impact on the model output. The key clinical features for predicting sepsis classification were determined based on the direction and importance of each clinical feature's impact on the model output.

6. The method for constructing a sepsis typing prediction model according to claim 5, wherein: The clinical characteristics included sex, age, body mass index, central venous oxygen saturation, peripheral perfusion index, temperature, heart rate, respiratory rate, mean arterial pressure, systolic blood pressure, diastolic blood pressure, lymphocyte percentage, mean corpuscular hemoglobin, prothrombin time, mean corpuscular hemoglobin concentration, indirect bilirubin measurement, albumin concentration, urea nitrogen concentration, platelet concentration, white blood cell count, hematocrit, fibrinogen content measurement, platelet distribution width, mean platelet volume, lymphocyte count, activated partial thromboplastin time measurement, hemoglobin concentration, absolute neutrophil count, hematocrit, SOFA total score, activated partial thromboplastin time ratio, vasoactive drug score, D-dimer concentration, creatinine concentration, alanine aminotransferase measurement, positive end-expiratory pressure, direct bilirubin measurement, total bilirubin measurement, and procalcitonin measurement.

7. The method for constructing a sepsis typing prediction model according to claim 6, wherein: The key clinical features refer to several clinical features ranked high in importance in the SHAP diagram corresponding to each clinical feature.

8. A method for applying a sepsis typing prediction model, characterized in that: include: Obtain the patient's clinical characteristics within the target time period; The clinical characteristics of the patient within the target time period are input into the sepsis classification prediction model to obtain the corresponding sepsis classification; the sepsis classification prediction model is obtained based on the method for constructing a sepsis classification prediction model based on dynamic blood flow and oxygen flow as described in any one of claims 3 to 7.

9. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the sepsis typing method according to any one of claims 1 to 2, or the method for constructing a sepsis typing prediction model according to any one of claims 3 to 7, or the method for applying a sepsis typing prediction model according to claim 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, it implements the sepsis typing classification method according to any one of claims 1 to 2, or the sepsis typing prediction model construction method according to any one of claims 3 to 7, or the sepsis typing prediction model application method according to claim 8.

Images