Systems and methods for ai based infection classification

Abstract

Methods, systems, and computer storage media may be used to detect and differentiate bacterial and viral infections. Cellular population parameters such as the monocyte distribution width (MDW) can be acquired for an individual. Clinical measurements such as a complete blood count (CBC), a measure of procalcitonin (PCT) or a measure of C-reactive protein (CRP)can be obtained for an individual. A trained model can be used to compute an output value from the parameters. The output value can be compared to a threshold value to determine a probability associated with a bacterial or viral infection. A report can be generated from the probability and sent to the appropriate party.

Description

SYSTEMS AND METHODS FOR Al BASED INFECTION CLASSIFICATIONCROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application is an international application which claims priority to and benefit of U.S. Provisional Application Serial No. 63 / 525,544 filed July 7, 2023, the contents of which are incorporated by reference in their entirety for all purposes.BACKGROUND

[0002] Differentiating bacterial and viral infections remains a challenge in the medical community. Often symptoms of a bacterial infection will parallel those of a viral infection and vice versa. Nevertheless, correctly diagnosing a bacterial or viral infection can be critical and sometimes even lifesaving to an individual. In the case of misdiagnosis of a bacterial infection as a virus can lead to a considerably delay in administering antibacterials. Erroneous diagnosis of this type can result in a medical emergency which could be lethal such as when an individual experiences a problematic case of sepsis. However, if the correct diagnosis is made in a timely fashion, a patient can be treated with empiric antibacterial therapy and generally quickly recovers. Conversely, misdiagnosis of a bacterial infection, when an individual’s symptoms are actually the result of a viral infection, can lead to complications as well. In such cases, a medical professional may erroneously prescribe an antibacterial when a viral infection is actually the root cause of the patient’s symptoms. An unnecessary antibacterial prescription can lead to loss and a disruption of the gut microflora which can result in opportunistic bacterial infections, such as Clostridioides difficile, to uncontrollably propagate causing painful inflammation and a potentially life-threatening situation. Additionally, unnecessary exposure can also lead to antimicrobial resistance within bacterial populations.

[0003] While methodologies to aid in correctly diagnosing bacterial and viral infections have been produced, development and advancements of technology in this area are needed to improve thequality, cost, and efficiency of medical care. Current approaches utilize procalcitonin (PCT), c- reactive protein (CRP), white blood cell (WBC), and neutrophil counts in a clinical setting to make early diagnosis of bacterial infections. However, studies suggest that these parameters do not reliably differentiate between a bacterial versus viral infection; meta-analysis of PCT in pneumonia found a 55% sensitivity and 76% specificity with AUROCs that vary from 0.60-0.85 (see PMID 31241140). Consequently, systems and methods that more accurately aid medical professionals in differentiating between bacterial and viral infections are desirable within the medical community.BRIEF SUMMARY

[0004] In various aspects, systems and methods are provided for training a model, and using the trained model to probabilistically assess an infection for an individual as more likely to be bacterial or more likely to be viral. The prediction probability can be derived by measuring and inputting parameters such as hematological cell parameters (e.g., cell population parameters), host biomarkers (e.g., proteins, RNAs, or metabolites), demographic information, and / or clinical measurements, including those that utilize hematological or metabolic parameters of an individual into a trained model. In aspects, parameters used as an input to the trained model may include a eosinophil count, neutrophil count, monocyte distribution width (MDW), red blood cell (RBC) count, basophil count, lymphocyte count, a measure of hemoglobin, white blood cell (WBC) count, red cell distribution width (RDW), monocyte count, a measure of hematocrit level, a measure of mean platelet volume (MPV), platelet count, monocyte cell population parameter, a granulocyte cell population parameter, a lymphocyte cell population parameter, a red blood cell population parameter, white blood cell count, platelet count, complete blood count (CBC), erythrocyte sedimentation rate (ESR), a neutrophil to lymphocyte ratio (NLR), eosinophil to lymphocyte ratio (ELR), lymphocyte to monocyte ratio (LMR), eosinophil-to-monocyte ratio (EMR), platelet-to-lymphocyte ratio (PLR), platelet-to-neutrophil ratio, monocyte to lymphocyte ratio (MLR), mean platelet volume-to-platelet count (MPV / PC) ratio, basophil to lymphocyte (BLR) ratio, anion gap, alkaline phosphatase (ALK), aspartate aminotransferase (AST), calcium, oxygen (O2) saturation, vimentin (VIM), TNF-related apoptosis-inducing ligand (TRAIL), procalcitonin (PCT), C Reactive Protein (CRP), interferon y- induced protein (IP- 10), Myxovirus resistance A (MxA), Cluster of Differentiation 64 (CD64), human neutrophile lipocalin (HNL), blood urea nitrogen (BUN), lactate, heparin-binding protein (HBP), adrenomedullin (ProADM), bioavailable adrenomedullin (bio-ARM), midregional adrenomedullin (MR-proADM), interleukin 6 (IL-6), ABL Proto-Oncogene 1 (ABL1), Interferon Regulatory Factor 9 (IRF9), Integrin Subunit Alpha M (ITGAM), Lymphocyte Antigen 6 Family Member E (LY6E), Proline-Serine-Threonine Phosphatase Interacting Protein 2 (PSTPIP2), runt-related transcription factor 1 (RUNX1), biological sex, height, weight, age, ethnicity, body mass index (BMI), medication history, primary language identifiers, clinician(s) responsible for patient care, known medical conditions, family history, diagnosis information, genetic information, patient complaints, and / or insurance information. In response to this input, an output value is generated. The output value may be compared to a threshold level to determine a prediction probability of a bacterial or viral infection. In other words, the parameters, through the use of a trained model, can be used to determine a predictive likelihood of a bacterial infection over a viral infection or vice versa.

[0005] In various aspects, a trained model is generated using a population of individuals with known bacterial or viral infections, of which a collection of parameters have been measured. The training model uses an array of parameters including hematological cell parameters, host-derived biomarkers, demographic information, and / or clinical measurements as an input which is processed through a model program to provide output data. The output data, when compared to a predicted output, is used to access the model. Through multiple rounds of iteration, the model is optimized until an acceptable level of predictability is obtained. The model can then be applied to parameters obtainedfrom an individual where a bacteria or viral infection is in question and a prediction probability can be determined.

[0006] In various aspects, the prediction probability obtained from the parameters, can be used as a way to aid medical professionals in generating a diagnosis for an individual. Similarly, the prediction probability could be used to help generate a treatment recommendation or a medical prescription for an individual. Additionally, or alternatively, the prediction probability can be used to help correct a treatment or medical prescription that was a produced erroneously. Additionally, or alternatively, the prediction probability can be used to recommend an additional diagnostic test.

[0007] In further aspects, the disclosed methods may be paired with an additional diagnostic tool to more accurately differentiate a bacterial and a viral infection. The additional diagnostic tool may be one or more of an assay that measures myxovirus resistance protein A (MxA), an assay that measures C-reactive protein (CRP); an assay that measures tumour necrosis factor-related apoptosisinducing ligand (TRAIL, an assay that measures interferon gamma induced protein- 10 (IP- 10), and combinations thereof. In aspects, the assay may be selected from PCR, RT-PCR, qPCR, ELISA, immunoassay, flow cytometry, or lateral flow assay an immunoassay. In further aspects, the additional diagnostic tool may be a bacterial culture test.BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS

[0008] To easily identify the discussion of any particular element or act, the most significant digit or digits in a reference number refer to the figure number in which that element is first introduced.

[0009] FIG. 1 illustrates a schematic depiction of an example operating environment, in accordance with aspects of the present disclosure;

[0010] FIG. 2 illustrates a schematic depiction of an example analyzer, in accordance with aspects of the present disclosure;

[0011] FIG. 3 illustrates a schematic depiction of an example analysis engine, in accordance with aspects of the present disclosure;

[0012] FIG. 4 illustrates a schematic depiction of a training model, in accordance with aspects of the present disclosure;

[0013] FIG. 5 illustrates a schematic depiction of providing input data into a trained model to generate an evaluation, in accordance with aspects of the present disclosure;

[0014] FIG. 6 illustrates a graph depiction of cell population parameter SHAP values for bacterial and viral infections, in accordance with aspects of the present disclosure;

[0015] FIG. 7 illustrates an aspect of the subject matter in accordance with one embodiment;

[0016] FIG. 8 illustrates a graph depiction of monocyte distribution width (MDW) for bacterial and viral infections, in accordance with aspects of the present disclosure;

[0017] FIG. 9 illustrates a method for generating a prediction probability for a bacterial or viral infection, in accordance with aspects of the present disclosure; and

[0018] FIG. 10 illustrates another method for generating a prediction probability for a bacterial or viral infection, in accordance with aspects of the present disclosure.DETAILED DESCRIPTION

[0019] The ability to differentiate between bacterial and viral infections, with a high degree of accuracy and in a timely manner, is desired by the medical community. The symptomatic presentation of bacterial or viral infections often contain considerable similarities making it challenging for medical clinicians to produce a correct diagnosis. Unfortunately, misdiagnosis of a viral or bacterial infection remains problematic. In some situations, bacterial infection may be incorrectly diagnosed as a viral infection, resulting in the prescription of antiviral treatment while failing to treat with an antibacterial. The untreated bacterial infection may lead to clinical complications, which could be life-threatening in certain cases. Conversely, in other circumstances a viral infection may be erroneously diagnosed as a bacterial infection. In this situation, antibacterial treatment would likely be ineffective as a form of treatment. Nevertheless, antibacterial use can result in detrimental effects to the individual, such as deregulation of the gut microflora ecosystem which can in itself result in a diseased state. Additionally, the over prescription of antibacterials can lead to populations of bacterial with antibiotic resistance thus diminishing their effectiveness over time. Correctly distinguishing bacterial and viral infections is paramount for the correct and effective treatment of individual and also has broader implications on a societal level by mitigating the emergence and spread of antimicrobial resistant pathogens.

[0020] Analyzing blood sample measurements for an individual is a common approach to differentiate between bacterial and viral infections. This approach traditionally involves quantifying levels of procalcitonin (PCT), c-reactive protein (CRP), white blood cell (WBC), and neutrophil counts. However, approaches that use only one parameter, such as PCT, CRP, or WBC alone often do not provide reliable prediction probabilities for distinguishing between a bacterial and viral infections. For example, recent meta-analysis in analyzing PCT for pneumonia patients resulted in a 55% sensitivity and 76% specificity with AUROCs that vary from 0.60-0.85 (see PMID 31241140). Such cases may result in a medical professional making an incorrect assessment of a viral infection when a bacterial infection is actually the cause of the patient’s symptoms.

[0021] As a means to improve predictive differentiation of bacterial and viral infections in a patient, systems, methods, processes, and media have been developed and are described herein. The systems, methods, processes, and media utilize and analyze one or more tests, obtained from an individual’s blood, serum, plasma, urine, other body fluids, or by other means, to deduce a prediction probability of a bacterial or virus infection using a trained model. In some aspects, the one or more tests can include the use of a detection method selected from one or more of PCR, RT-PCR, qPCR,ELISA, immunoassay, flow cytometry, hematology parameter signature, or lateral flow assay. In certain aspects, the one or more tests can include detection of one or more indicators or markers of an infection. For instance, in various aspects, the one or more tests can include the detection of one or more indicators of infection selected from eosinophil count, neutrophil count, monocyte distribution width (MDW), red blood cell (RBC) count, basophil count, lymphocyte count, a measure of hemoglobin, white blood cell (WBC) count, red cell distribution width (RDW), monocyte count, a measure of hematocrit level, a measure of mean platelet volume (MPV), platelet count, monocyte cell population parameter, a granulocyte cell population parameter, a lymphocyte cell population parameter, a red blood cell population parameter, white blood cell count, platelet count, complete blood count (CBC), erythrocyte sedimentation rate (ESR), a neutrophil to lymphocyte ratio (NLR), eosinophil to lymphocyte ratio (ELR), lymphocyte to monocyte ratio (LMR), eosinophil-to-monocyte ratio (EMR), platelet-to-lymphocyte ratio (PLR), platelet-to-neutrophil ratio, monocyte to lymphocyte ratio (MLR), mean platelet volume-to-platelet count (MPV / PC) ratio, basophil to lymphocyte (BLR) ratio, anion gap, alkaline phosphatase (ALK), aspartate aminotransferase (AST), calcium, oxygen (O2) saturation, vimentin (VIM), TNF-related apoptosis-inducing ligand (TRAIL), procalcitonin (PCT), C Reactive Protein (CRP), interferon y-induced protein (IP- 10), Myxovirus resistance A (MxA), Cluster of Differentiation 64 (CD64), human neutrophile lipocalin (HNL), blood urea nitrogen (BUN), lactate, heparin-binding protein (HBP), adrenomedullin (ProADM), bioavailable adrenomedullin (bio- ARM), midregional adrenomedullin (MR-proADM), interleukin 6 (IL-6), ABL Proto-Oncogene 1 (ABL1), Interferon Regulatory Factor 9 (IRF9), Integrin Subunit Alpha M (ITGAM), Lymphocyte Antigen 6 Family Member E (LY6E), Proline-Serine-Threonine Phosphatase Interacting Protein 2 (PSTPIP2), and runt-related transcription factor 1 (RUNX1).

[0022] In some aspects, the predictive differentiation is facilitated, at least in part, by artificial intelligence analysis of hematological cell parameters, clinical measurements, molecular parameters,patient demographic parameters, and any other categories or a combination of any of these types of parameters. Any one or combination of parameters may be reflective of a host response to a viral and / or bacterial infection. The output of the artificial intelligence analysis may be used by decision support tools incorporated in a provider-facing application, such as in an electronic health record application, to facilitate patient care decisions.

[0023] In an aspect, hematological cell parameters include one or more of cell population parameter, complete blood count (CBC), white blood cell count, ratios thereof, or a combination thereof. In certain aspects, cell population parameters can include eosinophil count, neutrophil count, monocyte distribution width (MDW), red blood cell (RBC) count, basophil count, lymphocyte count, a measure of hemoglobin, white blood cell (WBC) count, red cell distribution width (RDW), monocyte count, a measure of hematocrit level, a measure of mean platelet volume (MPV), platelet count, monocyte cell population parameter, a granulocyte cell population parameter, a lymphocyte cell population parameter, a red blood cell population parameter, white blood cell count, platelet count, complete blood count (CBC), a neutrophil to lymphocyte ratio (NLR), eosinophil to lymphocyte ratio (ELR), lymphocyte to monocyte ratio (LMR), eosinophil-to-monocyte ratio (EMR), platelet-to-lymphocyte ratio (PLR), platelet-to-neutrophil ratio, monocyte to lymphocyte ratio (MLR), mean platelet volume-to-platelet count (MPV / PC) ratio, basophil to lymphocyte (BLR) ratio and any other measurements that measure and / or compares the number, types, and / or properties of cells for an individual.

[0024] In an aspect, clinical measurements may be indicative of a host response and can include values representative of anion gap, a measure of procalcitonin (PCT), a measure of C-reactive protein (CRP), alkaline phosphatase (ALK), aspartate aminotransferase (AST), calcium, oxygen (O2) saturation, erythrocyte sedimentation rate (ESR), ratios thereof, and any other property that measuresthe concentration level or other properties of macromolecules, small molecules, or ions for an individual.

[0025] In an aspect, molecular parameters may be indicative of a host response and can include vimentin (VIM), TNF-related apoptosis-inducing ligand (TRAIL), procalcitonin (PCT), C Reactive Protein (CRP), interferon y-induced protein (IP-10), Myxovirus resistance A (MxA), Cluster of Differentiation 64 (CD64), human neutrophile lipocalin (HNL), blood urea nitrogen (BUN), lactate, heparin-binding protein (HBP), adrenomedullin (ProADM), bioavailable adrenomedullin (bio- ARM), midregional adrenomedullin (MR-proADM), interleukin 6 (IL-6), ABL Proto-Oncogene 1 (ABL1), Interferon Regulatory Factor 9 (IRF9), Integrin Subunit Alpha M (ITGAM), Lymphocyte Antigen 6 Family Member E (LY6E), Proline-Serine-Threonine Phosphatase Interacting Protein 2 (PSTPIP2), runt-related transcription factor 1 (RUNX1), and any other property that measures the concentration level or other properties of macromolecules (e.g., RNA or protein), small molecules, or ions for an individual. In various aspects, the one or more indicators of infection can be detected at the protein level and / or the mRNA level. In some examples, such a test may be a repeat of the initial test of specific samples as designated by the system in response to the results of the initial test, also known as reflex testing. For instance, tests for distinguishing a bacterial infection from a viral infection can include a) determining expression levels of RUNX1, ITGAM, PSTPIP2, LY6E, IRF9 and ABL1 in a biological sample from a patient; and b) determining that the patient has a viral infection when the expression level is greater than, less than, or equal to a first predetermined cutoff value or that the patient does has a bacterial infection when the expression level is greater than, less than, or equal to a second predetermined cutoff value. In some aspects, ABL1 can be used as a positive control biomarker that is indicative of the quality of the sample. In some aspects, determining a viral infection score can comprise inputting the expression levels of RUNX1, ITGAM, PSTPIP2, LY6E and IRF9 into an algorithm that determines the viral infection score. In some aspects, determining a bacterialinfection score can comprise inputting the expression levels of RUNX1, ITGAM, PSTPIP2, LY6E and IRF9 into an algorithm that determines the bacterial infection score. An algorithm that determines the viral infection score or bacterial infection score can be a product of a logistic regression model generated using the expression levels of RUNX1, ITGAM, PSTPIP2, LY6E, IRF9 and ABL1 measured in a training set of biological samples. The output of the artificial intelligence analysis may be used by decision support tools incorporated in an electronic health record application to facilitate patient care decisions.

[0026] Patient demographic parameters can include biological sex, height, weight, age, ethnicity, body mass index (BMI), medication history, primary language identifiers, clinician(s) responsible for patient care, known medical conditions, family history, diagnosis information, genetic information, patient complaints, insurance information or other physical characteristics and / or medical knowledge for an individual.

[0027] Turning to FIG. 1, FIG. 1 depicts an example system environment 100, in accordance with aspects herein. Example operating environment 100 includes one or more analyzer 102, one or more EHR server 104, at least one EHR database 106, a model execution environment 108, a user device 110, and a network 112.

[0028] The EHR database 106 stores patient information. For example, the patient information may include demographic parameters associated with one or more patients. The EHR database 106 may also include physiological parameters that are obtained from an instrument or instruments that measures cell population and / or molecular parameters. The database can be accessed and updated by authorized personnel from any remote computer with internet access. The EHR may also provide transfer and analysis of demographic parameters for individuals. Additionally, the EHR database may maintain diagnosis data for patient(s) in the form of ICD-9 and / or ICD-10 codes. The EHR database may also integrate with other healthcare software systems for seamless data transfer and analysis. Forexample, in some embodiments, the system may also include a laboratory information management system (LIMS) for managing laboratory workflows, inventory, and quality control. The LIMS may integrate with the EHR database and a hematological cell parameter, cell population parameter, clinical measurement, molecular parameter instrument(s), or any combination thereof, for streamlined data management and reporting.

[0029] Analyzer 102 is designed to perform various laboratory tests on patient samples. For example, the analyzer 102 may include an analyzer for measuring clinical parameters, an analyzer for measuring hematological cell parameters such as cell population parameters, an analyzer for measuring molecular parameters, or any other similar analytical instrumentation that is used to analyze patient samples. The analyzer 102 may use various techniques such as flow cytometry or immunodiagnostics to detect and quantify various blood, serum, plasma, urine, or other body fluid components such as red and white blood cells, platelets, hemoglobin, and antigen-antibody complexes. For example, the analyzer 102 can for measuring hematological cell parameters. Analyzers for clinical measurements such as PCT and CRP concentrations include a Cobas analyzer (Roche Diagnostics, Meylan, France), a Liaison XL (Diasorin, Saluggia, Italy), or a clinical chemistry analyzer such as AU5800 (Beckman Coulter, Inc, Brea, Calif., USA) or Olympus AU2700 plus analyzer (Beckman Coulter, Tokyo, Japan) analyzer, and can be used in the methods disclosed herein. The Beckman Coulter AU5800 Clinical Chemistry Analyzer carries out automated analysis of serum, plasma, urine samples and other body fluids and automatically generates results. The device is an automated chemistry analyzer that measures analytes in samples, in combination with appropriate reagents, calibrators, quality control (QC) material and other accessories. Applications include colorimetric, turbidimetric, latex agglutination, homogeneous enzyme immunoassay, and ion selective electrode (indirect potentiometry). Electrolyte measurement is performed using a single or double cell Ion Selective Electrode (ISE) which is also common amongthe other members of the AU family. Representative analytes that can be tested on the clinical chemistry analyzers include glucose, creatinine (Creat), urate, total bilirubin (Bil), cholesterol (Choi), tryglicerides (TG), calcium (Ca), phosphate (IP), iron (Fe), unsaturated iron binding capacity (UIBC), aspartate aminotransferase (AST), alanine aminotransferase (ALT), lactate dehydrogenase (LD), creatine kinase (CK), alpha-amylase (AMY), alkaline phosphatase (AP), potassium (K), sodium (Na), chloride (Cl), C-reactive protein (CRP), antistreptolysin O (ASO), rheumatoid factor (RF), and other clinical measurements as described herein. The instrument may also be equipped with an autosampling system for automated sample processing. Analyzer 102 communicates with the remote computer via a wired or wireless connection, allowing for real-time data transfer and monitoring. The instrument may also be equipped with software for automatic test result interpretation and quality control. The instrument may also include a built-in printer for printing test reports and labels.

[0030] The one or more user devices 110 are used to access and manage patient records stored in the EHR database. Authorized personnel can use the remote computer to schedule laboratory tests, view and interpret test results, and generate reports for patient records. The remote computer may also be equipped with software for data analysis and quality control. For example, the EHR server 104 may operate in a network 112 using logical connections to one or more analyzer 102, model execution environment 108, one or more user device 110, or any combination thereof. User device 110 may be located at a variety of locations in a medical or research environment, for example, but not limited to, clinical laboratories, hospitals and other inpatient settings, veterinary environments, ambulatory settings, medical billing and financial offices, hospital administration settings, home health-care environments, and clinicians’ offices. Clinicians may include, but are not limited to, a treating physician or physicians, specialists such as surgeons, radiologists, cardiologists, and oncologists, emergency medical technicians, physicians’ assistants, nurse practitioners, nurses, nurses’ aides, pharmacists, dieticians, microbiologists, laboratory experts, genetic counselors, researchers,veterinarians, students, and the like. The user device 110 may also be physically located in non- traditional medical care environments so that the entire health care community may be capable of integration on the network. The user device 110 may be personal computers, servers, routers, network PCs, peer devices, other common network nodes, or the like, and may include some or all of the components described above in relation to the EHR server 104. The devices can be personal digital assistants or other like devices.

[0031] EHR server 104 includes at least one EHR database 106. EHR database 106 may include structured data, unstructured data, or a combination of both. For example, EHR database 106 may include any and all information related, directly and / or indirectly, to patient care stored in an electronic health records (EHR) system of a health care provider, a network of health care providers, institutions, and / or a network of institutions. For instance, patient data may include, but is not limited to, an individual health record for each patient, metadata, and appointment scheduling information. A health record may contain physiological information such as hematological cell parameters such as cell population parameters, clinical measurements, molecular parameters, and demographic parameters. Further, the health record may contain similar data recorded at various times such as: previous appointments; emergency room visits; surgeries; therapy sessions; urgent care visits; and so on. Additionally, the health record may contain data recorded in other EHR systems at other health care providers, health care provider networks, institutions, and / or institution networks. This data can be stored in a variety of different formats such as unstructured text, structured text, images, buttons, automatically populated fields, graphs, and so on.

[0032] Model execution environment 108 includes hardware, software, firmware, or any combination thereof that work together to support the development, training, and deployment of analysis engine 300. For example, some aspects of model execution environment 108 include one or more servers or clusters of servers that host one or more instances of the analysis engine. The serversmay include specialized hardware such as graphics processing units (GPUs) or field-programmable gate arrays (FPGAs) that are optimized for machine learning workloads. In an embodiment, the one or more servers or clusters of servers may comprise EHR server 104 or one or more servers or clusters of servers communicatively coupled to the EHR server 104. In another aspect, model execution environment 108 is integrated with analyzer 102. For example, the Al model may be stored within the local storage of analyzer 102. Further, some embodiments of analyzer 102 may include specialized hardware such as GPUs that are optimized for machine learning workloads.

[0033] Network 112 may include, without limitation, local area networks (LANs) and / or wide area networks (WANs). Such networking environments are commonplace in offices, enterprise-wide computer networks, intranets, and the Internet. When utilized in a WAN networking environment, the server 102 may include a modem or other means for establishing communications over the WAN, such as the Internet. In a networked environment, program modules or portions thereof may be stored in the EHR server 104, in the database cluster 104, on any of the remote computers 108, or one any of the user devices 110. For example, and not by way of limitation, various application programs may reside on the memory associated with any one or more of the remote computers 108 and / or any one or more of the user devices 110. It will be appreciated by those of ordinary skill in the art that the network connections shown are exemplary and other means of establishing a communications link between the computers (e.g., server 102 and remote computers 108) may be utilized.

[0034] With general reference to FIG. 2, there are many thousands of possible combinations of sensor readings and calculated relationships that might correlate to a particular characteristic of a blood sample, and, once subpopulations of cells have been identified, a particular subpopulation of cells may be further characterized by one or more sensor readings, such as light scatter (e.g., LALS, ALL, UMALS, LMALS, and MALS), impedance, volume, conductivity, light scatter turbidity, mean, standard deviation of volume, conductivity, or any combination thereof. These sensor reading-basedcharacterizations may be made in addition to or in lieu of cytochemical staining, marker affinity, or other cell identification techniques. That is, hematology analyzers (e.g., hematology analyzer 200) can often provide data about a subpopulation of cells that is much richer than simply a count or proportion of those cells compared to other subpopulations of cells within a sample. One example is Monocyte Distribution Width (MDW), a calculation of the standard deviation of cell volumes within the subpopulation of monocytes within a blood sample. This characterization of the monocyte population is associated with sepsis, as described, for example, in US patent 11,114,205 and US patent applications 17 / 391,599; 16 / 390,648; 16 / 925,943; 16 / 925,937; and 16 / 390,597. In some cases, more than one characterization of a subpopulation of cells or relationship between subpopulation of cells may be indicative of the same or related conditions, such as viral infection, sepsis, anemia, leukemia, etc.

[0035] Beckman Coulter’ s DxH hematology analyzers (including but not limited to the DxH 800 and DxH 900) perform at least seven parameter measurements that characterize cellular size, shape and morphology. The seven parameters - impedance, radio frequency, and five laser light scatter measurements - are captured simultaneously for each cell passing through the flow cell. These parameters generally measure volume, conductivity, and scatter (VCS). In addition, each parameter measurement includes four pulse shape attributes. These measurements plus time provide a total of 29 distinct measurements per cellular event. The impedance measurement (volume) is an indicator of three-dimensional cell size. The radio frequency measurement (conductivity) provides information regarding the internal structure of the cell, such as cellular density. The DxH systems use high-speed, high-resolution analog-to-digital conversion with Digital Signal Processing (DSP) circuitry to measure multiple parameters for each cellular event. DSP algorithms analyze the cellular data digitally, providing cellular definition and resolution.Differential accuracy and flagging technology are obtained by combining the additional lightscatter measurements with data analysis techniques to further define and separate cell populations.

[0036] Other hematology analyzers, such as those manufactured by Abbott, Sysmex, and Mindray, use the cluster of differentiation (CD) system to differentiate white blood cell populations. CD molecules can act in numerous ways, often acting as receptors or ligands, by which a signal cascade is initiated, altering the behavior of the cell. Some CD proteins do not play a role in cell signaling, but have other functions, such as cell adhesion. The CD system nomenclature commonly used to identify cell markers thus allows cells to be defined based on what molecules are present on their surface. There are more than 350 CD molecules identified for humans. For example, monocytes can be identified with CD45+ and CD14+. Using fluorescently labeled antibodies, these cell markers can used to sort cells. Fluorescence activated cell sorting (FACS) provides a method of sorting a heterogeneous mixture of cells into two or more containers, a single cell at a time, based upon the specific light scattering and fluorescent characteristics of each cell. Fluorescence flow cytometry or FACS may also provide information about cell composition of a labeled cell. For example, information about cell density or complexity may be obtained by measuring light scattered by the cell and information about cell size and internal structure may be obtained by measuring the fluorescence signal intensity of the cell. Systems based on fluorescence flow cytometry or FACS report “FACS parameters.” VCS parameters and FACS parameters are both described in the art and both may be used in methods or systems implemented based on this disclosure.

[0037] With specific reference to FIG. 2, an example hematology analyzer 200 is depicted in accordance with aspects described herein. In some aspects, hematology analyzer 200 may be an analyzer 102 of FIG. 1. As shown here, hematology analyzer 200 includes a transducer module 208 having a light or irradiation source such as a laser 210 emitting a beam 212. The laser 210 can be, for example, a 635 nm, 5 mW, solid-state laser. In some instances, hematology analyzer 200 may includea focus-alignment system 218 that adjusts beam 212 such that a resulting beam 220 is focused and positioned at a cell interrogation zone 230 of a flow cell 230. In some instances, the flow cell 230 receives a sample aliquot from a preparation system 202. Various fluidic mechanisms and techniques can be employed for hydrodynamic focusing of the sample aliquot within flow cell 230.

[0038] In some instances, the aliquot generally flows through the cell interrogation zone 230 such that its constituents pass through the cell interrogation zone 230 one at a time. In some cases, a hematology analyzer 200 may include a cell interrogation zone or other feature of a transducer module or blood analysis instrument such as those described in U.S. Pat. Nos. 5,125,737; 6,226,652; 7,390,662; 8,094,299; 8,189,187; and 9,939,453, the contents of which are incorporated herein by reference for all purposes. For example, a cell interrogation zone 230 may be defined by a square transverse cross-section measuring approximately 50x50 microns, and having a length (measured in the direction of flow) of approximately 65 microns. Flow cell 230 may include an electrode assembly having first and second electrodes 234, 236 for performing DC impedance and / or RF conductivity measurements of the cells passing through cell interrogation zone 230. Signals from electrodes 234, 236 can be transmitted to the analysis system 202. The electrode assembly can analyze volume and conductivity characteristics of the cells using low-frequency current and high-frequency current, respectively. For example, low-frequency DC impedance measurements can be used to analyze the volume of each individual cell passing through the cell interrogation zone. High-frequency RF current measurements can be used to determine the conductivity of cells passing through the cell interrogation zone. Because cell walls act as conductors to high frequency current, the high frequency current can be used to detect differences in the insulating properties of the cell components, as the current passes through the cell walls and through each cell interior. High frequency current can be used to characterize nuclear and granular constituents and the chemical composition of the cell interior.

[0039] The light source in FIG. 2 has been described as a laser, however, the light source may alternatively or additionally include a xenon lamp, an LED lamp, an incandescent lamp, or any other suitable source of light, including combinations of the same or different kinds of lamps (e.g., multiple LED lamps or at least one LED lamp and at least one xenon lamp). As shown in FIG. 2, for example, incoming beam 220 irradiates the cells passing through cell interrogation zone 230, resulting in light propagation within an angular range a (e.g. scatter, transmission) emanating from the zone 230. Exemplary systems are equipped with sensor assemblies that can detect light within one, two, three, four, five, or more angular ranges within the angular range a, including light associated with an extinction or axial light loss measure. As shown, light propagation 240 can be detected by a light detection assembly 250, optionally having a light scatter detector unit 250A and a light scatter and / or transmission detector unit 250B. In some instances, light scatter detector unit 250A includes a photoactive region or sensor zone for detecting and measuring upper median angle light scatter (UMALS), for example, light that is scattered or otherwise propagated at angles relative to a light beam axis within a range from about 20 to about 42 degrees. In some instances, UMALS corresponds to light propagated within an angular range from between about 20 to about 43 degrees, relative to the incoming beam axis, which irradiates cells flowing through the interrogation zone. Light scatter detector unit 250A may also include a photoactive region or sensor zone for detecting and measuring lower median angle light scatter (LMALS), for example, light that is scattered or otherwise propagated at angles relative to a light beam axis within a range from about 10 to about 20 degrees. In some instances, LMALS corresponds to light propagated within an angular range from between about 9 to about 19 degrees, relative to the incoming beam axis which irradiates cells flowing through the interrogation zone.

[0040] A combination of UMALS and LMALS is defined as median angle light scatter(MALS), which may be light scatter or propagation at angles between about 9 degrees and about 43degrees relative to the incoming beam axis which irradiates cells flowing through the interrogation zone. One of skill in the art will understand that these angles (and the other angles described herein) may vary somewhat based on the configuration of the interrogation, sensing and analysis systems.

[0041] As shown in FIG. 2, the light scatter detector unit 250A may include an opening 251 that allows low angle light scatter or propagation 240 to pass beyond light scatter detector unit 250A and thereby reach and be detected by light scatter and transmission detector unit 250B. According to some embodiments, light scatter and transmission detector unit 250B may include a photoactive region or sensor zone for detecting and measuring lower angle light scatter (LALS), for example, light that is scattered or propagated at angles relative to an irradiating light beam axis of less than about 5.1 degrees. In some instances, LALS corresponds to light propagated at an angle of less than about 9 degrees, relative to the incoming beam axis, which irradiates cells flowing through the interrogation zone. In some instances, LALS corresponds to light propagated at an angle of less than about 10 degrees, relative to the incoming beam axis, which irradiates cells flowing through the interrogation zone. In some instances, LALS corresponds to light propagated at an angle of about 1.9 degrees±0.5 degrees, relative to the incoming beam axis, which irradiates cells flowing through the interrogation zone. In some instances, LALS corresponds to light propagated at an angle of about 3.0 degrees±0.5 degrees, relative to the incoming beam axis, which irradiates cells flowing through the interrogation zone. In some instances, LALS corresponds to light propagated at an angle of about 3.7 degrees±0.5 degrees, relative to the incoming beam axis, which irradiates cells flowing through the interrogation zone. In some instances, LALS corresponds to light propagated at an angle of about 5.1 degrees±0.5 degrees, relative to the incoming beam axis, which irradiates cells flowing through the interrogation zone. In some instances, LALS corresponds to light propagated at an angle of about 7.0 degrees±0.5 degrees, relative to the incoming beam axis, which irradiates cells flowing through the interrogation zone. In each instance, LALS may correspond to light propagated an angle of about 1.0 degrees ormore. That is, LALs may correspond to light propagated at angles between about 1.0 degrees and about 1.9 degrees; between about 1.0 degrees and about 3.0 degrees; between about 1.0 degrees and about 3.7 degrees; between about 1.0 degrees and about 5.1 degrees, between about 1.0 degrees and about 7.0 degrees, between about 1.0 degrees and about 9.0 degrees; or between about 1.0 degrees and about 10.0 degrees.

[0042] According to some embodiments, light scatter and transmission detector unit 250B may include a photoactive region or sensor zone for detecting and measuring light transmitted axially through the cells, or propagated from the irradiated cells, at an angle of about 0 degrees relative to the incoming light beam axis. In some cases, the photoactive region or sensor zone may detect and measure light propagated axially from cells at angles of less than about 1 degree relative to the incoming light beam axis. In some cases, the photoactive region or sensor zone may detect and measure light propagated axially from cells at angles of less than about 0.5 degrees relative to the incoming light beam axis less. Such axially transmitted or propagated light measurements correspond to axial light loss (ALL or AL2). As noted in previously incorporated U.S. Pat. No. 7,390,662, when light interacts with a particle, some of the incident light changes direction through the scattering process (i.e., light scatter) and part of the light is absorbed by the particles. Both of these processes remove energy from the incident beam. When viewed along the incident axis of the beam, the light loss can be referred to as forward extinction or axial light loss. Additional aspects of axial light loss measurement techniques are described in U.S. Pat. No. 7,390,662 at column 5, line 58 to column 6, line 4.

[0043] As such, the hematology analyzer 200 provides means for obtaining light propagation measurements, including light scatter and / or light transmission, for light emanating from the irradiated cells of the biological sample at any of a variety of angles or within any of a variety of angular ranges, including ALL and multiple distinct light scatter or propagation angles. For example, light detectionassembly 250, including appropriate circuitry and / or processing units, provides a means for detecting and measuring UMALS, LMALS, LALS, MALS, and ALL.

[0044] Wires or other transmission or connectivity mechanisms can transmit signals from the electrode assembly (e.g. electrodes 234, 236), light scatter detector unit 250A, and / or light scatter and transmission detector unit 250B to the analysis system 202 for processing. For example, measured DC impedance, RF conductivity, light transmission, and / or light scatter parameters can be provided or transmitted to the analysis system 202 for data processing. In some instances, analysis system 202 may include computer processing features and / or one or more modules or components, which can evaluate the measured parameters, identify and enumerate biological sample constituents, and correlate a subset of data characterizing elements of the biological sample with one or more features or parameters of interest. Some aspects of analysis system 202 include an analysis engine such as described in relation to FIG. 3.

[0045] Additionally, or alternatively, as depicted in FIG. 2, hematology analyzer 200 may generate or output a report 204 presenting measurements made or parameters calculated for the sample. The measurements made or parameters calculated for a sample can include UMALS, LMALS, LALS, MALS, ALL, WBC, MDW, monocyte %, absolute lymphocyte count (ALC), lymphocyte %, eosinophil %, absolute neutrophil count (ANC), neutrophil %, or any combination thereof.

[0046] In some instances, excess biological sample from transducer module 208 can be directed to an external (or alternatively internal) waste system 208. In some instances, the hematology analyzer 200 may include one or more features of a transducer module or blood analysis instrument such as those described in previously incorporated U.S. Pat. Nos. 5,125,737; 6,226,652; 8,094,299; 8,189,187 and 9,939,453.

[0047] FIG. 3 depicts an example analysis engine 300, in accordance with aspects described herein. Aspects of analysis engine 300 can be incorporated into a processing feature and / or module or component of an analyzer (such as depicted in FIG. 2), an application executed by a remote device (e.g., user device 110 depicted in FIG. 1), or can operate as an independent component of an operating environment (e.g., model execution environment 108 depicted in FIG. 1).

[0048] Generally, the analysis engine 300 evaluates a set of measurements or parameters, identifies and enumerates biological sample constituents, and correlates a subset of data characterizing elements of the biological sample with one or more features or parameters of interest. The analysis engine 300 comprises a receiver 302, a model engine 304, a decision rules analyzer 306, and a communicator 308.

[0049] A receiver, such as receiver 302, generally collects measurements made or parameters calculated based on analysis of an individual’s sample. The data (e.g., measurements made or parameters calculated) can be received directly from a subsystem of an analyzer or from a data store in some aspects. Receiver 302 can use any data collection technique known in the art.

[0050] Model engine 304 includes modules that include logical expressions for the evaluation of measurements and parameters received by the analysis engine 300. The logical expressions can include linear or parallel processes that evaluate measurements made by or parameters calculated by an analyzer, such as analyzer 102 described in relation to FIG 1 or hematology analyzer 200 described in relation to FIG. 2. Further, model engine 304 comprises a library of rules, models, and logic expressions in any combination that generate an output value or set of output values which further analyzed by the decision rules analyzer. The models within the acuity analyzer are capable of being trained using a dataset of individuals with known bacterial or viral infections, as discussed in FIG. 4.

[0051] Decision rules analyzer 306 comprises a library of decision rules. A decision rule is a logic expression that compares an individual parameter or characteristic of a blood, serum, plasma,urine, or other body fluid sample with a threshold value. The decision rules analyzer 306 assembles one or more decision rules from the library to build a logical expression that the analysis engine can evaluate. In one aspect, the decision rules analyzer 306 can utilize a linear combination or two or more parameters. In combination, the decision rules can be used to determine a probability that an individual associated with a blood, serum, plasma, urine, or other body fluid sample has a bacterial or viral infection. A potential outcome can be associated with a recommendation, treatment, or intervention in some aspects. For example, where an individual’s outcome corresponds with a risk of a bacterial infection, a recommendation to initiate antibacterial therapy (e.g., antibiotics therapy) can be associated therewith. Additionally, in some embodiments, decision rules analyzer 306 can include rules, models, logic expressions, in any combination that are configured to forecast medical conditions.

[0052] In some aspects, the data analysis engine 300 can incorporate the operations of one or more analyzer modules to generate an output. For example, the decision rules maintained by a decision rules analyzer 306 can be used to determine if an individual is likely to have a bacterial or viral infection. In response to determination that the individual has a probability of infection above a particular threshold, some aspects of the data analysis engine 300 can activate the decision rules analyzer 306 to facilitate the determining if the individual is at an elevated risk of medical complications arising from the infection.

[0053] Communicator 308 generally communicates the results of the analysis engine 300 to at least one predetermined target. In some aspects, the predetermined target can include a remote device that is executing a local client of a laboratory information system or a local client of an electronic medical record system (e g., user device 110 described in relation to FIG. 1). In such aspects, the results can include presentation of a visual display or audio signal that provides a probability of a bacterial or viral infection, a recommendation of diagnosis, notification of a potentially erroneoustreatment recommendation or medical prescription, or an alert that the individual corresponding to the analyzed sample may develop sepsis or other severe condition. In other aspects, the results can be a display of a treatment recommendation, an additional diagnostic test recommendation, a triage recommendation, a prognosis, or a medical prescription. In other aspects, the results can be used to generate a recommendation for an additional diagnostic test that can further distinguish between a bacterial or a viral infection. The additional diagnostic test to distinguish between bacterial infection or viral infection may be selected from one or more of, PCR, RT-PCR, qPCR, ELISA, immunoassay, flow cytometry, or lateral flow assay. In aspects, the additional diagnostic test to distinguish between bacterial infection or viral infection comprises detecting host-derived protein biomarkers in response to a viral or bacterial infection. In aspects, the additional diagnostic test to distinguish between bacterial infection or viral infection comprises detecting one or both of a host gene expression response to a viral or bacterial infection. In aspects, the additional diagnostic test to distinguish between bacterial infection or viral infection comprises detecting one or more indicators of infection selected from vimentin (VIM), TNF-related apoptosis-inducing ligand (TRAIL), procalcitonin (PCT), C Reactive Protein (CRP), interferon y-induced protein (IP-10), Myxovirus resistance A (MxA), Cluster of Differentiation 64 (CD64), human neutrophile lipocalin (HNL), blood urea nitrogen (BUN), lactate, heparin-binding protein (HBP), pro-adrenomedullin (ProADM), interleukin 6 (IL-6), ABL Proto-Oncogene 1 (ABLl), Interferon Regulatory Factor 9 (IRF9), Integrin Subunit Alpha M (ITGAM), Lymphocyte Antigen 6 Family Member E (LY6E), Proline-Serine-Threonine Phosphatase Interacting Protein 2 (PSTPIP2), and runt-related transcription factor 1 (RUNX1). In aspects, the additional diagnostic test to distinguish between bacterial infection or viral infection comprises determining expression levels of RUNX1, ITGAM, PSTPIP2, LY6E, and IRF9. The additional diagnostic tool may comprise detection of a host gene expression response that discriminates between a bacterial infection and a viral infection based on detecting the expression level of a combination ofABL1, IRF9, ITGAM, LY6E, PSTPIP2 and RUNX1 in biological samples from the human subject and determining whether the human subj ect has a bacterial or viral infection based on those expression levels, as described in, for example, WO2022197351. In brief, the detection of a host gene expression response may comprise: determining if a subject has a bacterial infection, a viral infection, or a non- infectious cause of fever, further comprising: a) determining the expression levels of RUNX1, ITGAM, PSTPIP2, LY6E, IRF9 and ABL1 in a biological sample from the subject; b) determining that the expression level of ABL1 is greater than a first predetermined cutoff value; c) determining a viral infection score and a bacterial infection score based on the expression levels of RUNX1, ITGAM, PSTPIP2, LY6E and IRF9; d) comparing the viral infection score to a second predetermined cutoff value and the bacterial infection score to a third predetermined cutoff value; and e) determining: that the subject has a viral infection when the viral infection score is greater than the second predetermined cutoff value or that the subject does not have viral infection when the viral infection score is less than the second predetermined cutoff value, that the subject has a bacterial infection when the bacterial infection score is greater than the third predetermined cutoff value or the subject does not have a bacterial infection when the bacterial infection score is less than or equal to the third predetermined cutoff value, and that the subject has a non-infectious cause of fever in the subject when the viral infection score is less than the second predetermined cutoff value and the bacterial infection score is less than the third predetermined cutoff value. In further aspects, the additional diagnostic test may be a single biomarker test, such as CRP or PCT. In further aspects, the additional diagnostic test may employ a a protein panel, for example, PCT+IP-10+CRP or CRP + MxA. In further aspects, the additional diagnostic test may be a gene expression panel, such as a 10-Gene signature panel (as described in Bhattacharya et al, Sci Rep. 2017; 7:6548) or 130-Gene signature panel (as described in Tsalik et al, Sci Transl Med. 2016; 8: 322ral 1). In further aspects, the additional diagnostic test may be a test administered via a wearable technology, for example an ADAMM-RSM or Fitbit.

[0054] In some aspects, the predetermined target can include a data store maintaining a laboratory information system or an electronic medical record system (e.g., EHR database 106 described in relation to FIG. 1).

[0055] With reference to FIG. 4, FIG. 4 is an example data flow illustrating an example process 400 for training one or more Al model(s) to generate a prediction probability that an individual as being more likely to be bacteria or more likely to be viral based on the output of an analyzer, in accordance with some embodiments of the present disclosure. It should be understood that this and other arrangements described herein are set forth only as examples. Other arrangements and elements (e.g., machines, interfaces, functions, orders, groupings of functions, etc.) may be used in addition to or instead of those shown, and some elements may be omitted altogether. Further, many of the elements described herein are functional entities that may be implemented as discrete or distributed components or in conjunction with other components, and in any suitable combination and location. Various functions described herein as being performed by entities may be carried out by hardware, firmware, and / or software.

[0056] At a high level, the process 400 may include one or more Al model(s) 404 receiving inputs, such as analyzer data 402, and generating one or more outputs, such as output data 406. As used in reference to training a model, the analyzer data 402 may be referred to as training data. Analyzer data 402 may include the output of one or more analyzers 102. For example, an analyzer may output cell population parameters (e g., eosinophil count, neutrophil count, MDW, RBC count, basophil count, lymphocyte count, a measure of hemoglobin, WBC count, RDW, monocyte count, a measure of hematocrit level, a measure of MPV, platelet count, or a combination thereof), ratios thereof (e.g., neutrophil to lymphocyte ratio (NLR), eosinophil to lymphocyte ratio (ELR), lymphocyte to monocyte ratio (LMR), eosinophil-to-monocyte ratio (EMR), platelet-to-lymphocyte ratio (PLR), platelet-to-neutrophil ratio, monocyte to lymphocyte ratio (MLR), mean platelet volume-to-platelet count (MPV / PC) ratio, basophil to lymphocyte (BLR) ratio, and so forth), clinical measurements (e.g., erythrocyte sedimentation rate (ESR), anion gap, a measure of procalcitonin (PCT), a measure of C-reactive protein (CRP), a measure of alkaline phosphatase (ALK), a measure of aspartate aminotransferase (AST), a measure of calcium, or oxygen (O2) saturation), molecular parameters (e.g., protein or mRNA levels of vimentin (VIM), TNF-related apoptosis-inducing ligand (TRAIL), procalcitonin (PCT), C Reactive Protein (CRP), interferon y-induced protein (IP- 10), Myxovirus resistance A (MxA), Cluster of Differentiation 64 (CD64), human neutrophile lipocalin (HNL), blood urea nitrogen (BUN), lactate, heparin-binding protein (HBP), adrenomedullin (ProADM), bioavailable adrenomedullin (bio-ARM), midregional adrenomedullin (MR-proADM), interleukin 6 (IL-6), ABL Proto-Oncogene 1 (ABL1), Interferon Regulatory Factor 9 (IRF9), Integrin Subunit Alpha M (ITGAM), Lymphocyte Antigen 6 Family Member E (LY6E), Proline-Serine- Threonine Phosphatase Interacting Protein 2 (PSTPIP2), runt-related transcription factor 1 (RUNX1), and so forth), or any combination thereof.

[0057] Process 400 may include generating and / or receiving analyzer data 402 from one or more analyzers 102 (e.g., hematology analyzer 200 of FIG. 2 and / or a clinical chemistry analyzer such as AU5800), EHR database 106, or a combination there of. The analyzer data 402 may be received, as a non-limiting example, from one or more sensors of a machine (e.g., analyzer 102 of FIG. 1). The analyzer data 402 may be used by hematology analyzer 200, and within process 400, to identify cell population parameters, hematological cell parameters, or a combination thereof. The analyzer data 402 may include, without limitation, analyzer data 402 from any of the sensors of the analyzer 102 vehicle including, for example and with reference to FIG. 2, flow cell 230, electrodes 234, 236 light scatter detector unit 250A, transmission detector unit 250B, and / or other sensor types. For example, analyzer data 402 may include output from sensors of a PCR, RT-PCR, qPCR, ELISA, immunoassay, flow cytometry, colorimetric, turbidimetric, latex agglutination, ion selective electrode, or lateral flowassay. Furthermore, the process 400 can be extended to include other types of laboratory tests, such as imaging studies and microbiological cultures. These additional tests can provide valuable information that can aid in the diagnosis of bacterial infections. For example, chest X-rays and computed tomography (CT) scans can be used to identify signs of pneumonia, which is often caused by bacterial infections. Microbiological cultures can also be used to identify the specific bacteria causing an infection, which can inform treatment decisions. In other aspects, the additional test may be an additional diagnostic test for distinguishing between a bacterial and viral infection, for example, a host response-based diagnostic test that detects, in the host, procalcitonin (PCT) erythrocyte sedimentation rate (ESR), IL-4, IL-5, IL-6, IL- 10, granulocyte-macrophage colony-stimulating factor, lactic acid, pyroglutamic acid, CD64, L-lactate, soluble triggering receptor expressed on myeloid cells 1 (sTREM-1), mid-regional pro-adrenomedullin (MR-proADM), metalloproteinase-9, a three-protein signature of TRAIL, interferon gamma-induced protein 10 (IP- 10), and CRP (TRAIL / 1P-10 / CRP), and combinations thereof. In other aspects, the additional test may comprise data obtained via a wearable device, such as via a Fitbit or ADAMM-RSM device. In further aspects, the additional test may comprise gene expression data obtained from the host.

[0058] The Al model(s) 404 may be trained using the analyzer data 402 as well as corresponding ground truth data (e.g., ground truth data 408). The ground truth data may include annotations, labels, treatments (e.g., prescriptions, therapies, care unit transfers, other EHR database data and so forth), diagnosis at discharge, and / or the like. For example, in some aspects, the ground truth data may include cell population parameters, ratios thereof, or any combination thereof at multiple time points throughout treatment of an individual. Additionally, in some aspects, the ground truth data may include diagnosis (e.g., ICD-9 code), treatment data, or demographic data for the individual.

[0059] Al model(s) 404 may include any type of machine learning model, such as a machine learning model(s) using linear regression, logistic regression, decision trees, support vector machines (SVM), Naive Bayes, k-nearest neighbor (Knn), K means clustering, random forest, dimensionality reduction algorithms, gradient boosting algorithms, neural networks (e.g., auto-encoders, convolutional, recurrent, perceptrons, long / short term memory / LSTM, Hopfield, Boltzmann, deep belief, deconvolutional, generative adversarial, liquid state machine, etc.), computer vision algorithms, and / or other types of machine learning models.

[0060] Al model(s) 404 may include any number of layers. One or more of the layers may include an input layer. The input layer may hold values associated with the analyzer data 402. One or more layers may include convolutional layers. The convolutional layers may compute the output of neurons that are connected to local regions in an input layer, each neuron computing a dot product between their weights and a small region they are connected to in the input volume. One or more of the layers may include one or more fully connected layer(s). Each neuron in the fully connected layer(s) may be connected to each of the neurons in the previous volume. The fully connected layer may compute class scores, and the resulting volume may be 1 x 1 x number of classes. In some nonlimiting aspects, the Al model(s) 404 may include a series of convolutional and max pooling layers to facilitate analyzer data 402 feature extraction, followed by multi-scale convolutional and up- sampling layers to facilitate global context feature extraction. In addition, some of the layers may include parameters (e.g., weights and / or biases), such as the convolutional layers and the fully connected layers, while others may not, such as the pooling layers.

[0061] With reference to the annotations 410 may be generated within a labeling program, another type of program suitable for generating the annotations 410 or other of the ground truth data, and / or may be done by manual retrospective analysis of EHR records, in some examples. In any example, the annotations 410, and / or the ground truth data 408 may be synthetically produced (e.g.,generated from computer models or renderings), real produced (e.g., designed and produced from real-world data), machine-automated (e.g., using natural language detection to extract features from analyzer data 402 and then generate labels), human annotated (e.g., labeler, or annotation expert, defines the labels), and / or a combination thereof. An encoder may use the annotations to encode the ground truth data.

[0062] The ground truth data 408 and annotations 410 may be combined, in some aspects, to generate predicted output 412. Said differently, each set of analyzer data 402 associated with a specific individual may be combined with the corresponding annotations 410 to generate a predicted output 412 for the set of analyzer data 402. Predicted outputs 412 may include the computational equivalent of one of two options (e g., viral infection or bacterial infection) in some aspects. Alternatively, predicted outputs 412 may include the computational equivalent of one of three options (e.g., viral infection, bacterial infection, or both).

[0063] Processes 400 includes one or more loss functions 414. In some aspects, loss function 414 is biased loss function that prioritizes accurately identifying true positive bacterial infections and minimizing false positive bacterial infections. This may be achieved by assigning a higher weight to false positive bacterial infections than false positive viral infections in the loss function. The resultant comparison of output data 406 and 412 may be evaluated by loss function 414 and trained using a reinforcement learning algorithm. The reinforcement learning algorithm rewards the Al for correctly identifying bacterial infections and penalizes it for falsely identifying viral infections as bacterial infections. This further reinforces the Al's ability to accurately distinguish between viral and bacterial infections.

[0064] The Al model(s) 404 are trained using various machine learning algorithms, including neural networks, decision trees, or support vector machines, among others. The Al is trained to learn the relationship between the laboratory test variables and the type of infection (viral or bacterial)present in the patient. The Al is then validated using the validation set to ensure that it is able to accurately distinguish between viral and bacterial infections.

[0065] In one embodiment, the Al model(s) 404 is further trained using a reinforcement learning algorithm. The reinforcement learning algorithm rewards the Al for correctly identifying bacterial infections and penalizes it for falsely identifying viral infections as bacterial infections. This further reinforces the Al's ability to accurately distinguish between viral and bacterial infections.

[0066] Once the Al is trained and validated, it can be integrated into a clinical decision-making tool, such as a clinical decision support system, to aid in the diagnosis of bacterial infections in patients. In aspects, once the Al is trained and validated, it can be integrated into a clinical decisionmaking tool to aid in the diagnosis of bacterial infections in patients. In aspects, once the Al is trained and validated, it can be integrated into a clinical decision-making tool for the purpose of patient triaging. In aspects, once the Al is trained and validated, it can be used to support therapeutic decision making. In aspects, once the Al is trained and validated, it can be used to estimate prognosis (including critical care outcome, emergent procedure, hospitalization, and combinations thereof. The Al model can be used as a clinical decision support system to analyze patient data to identify patterns indicative of a bacterial infection, aiding in patient triaging by prioritizing based on the severity of the condition. In further aspects, the Al model can be used as a clinical decision support system to suggest effective treatment options by considering the patient’s overall health, potential drug interactions, and the type of bacteria or virus involved. In further aspects, the Al model can be used as a clinical decision support system to estimate outcomes such as the need for critical care, emergency procedures, or hospitalization based on factors such as the severity of the infection, the patient’s response to treatment, their overall health status, or other clinical parameters as described herein. The process 400 can be customized to meet the specific needs of different healthcare organizations and patient populations. For example, the system can be adapted to work with different types of electronic healthrecord databases, or to incorporate different laboratory tests and diagnostic algorithms. Accordingly, Al model(s) 404 can be used to support medical professionals in making more accurate diagnoses, reducing the likelihood of antibiotic overuse and underuse, and ultimately improving patient outcomes.

[0067] For example, in using the model for training one or more Al model(s) to generate a prediction probability described in FIG 4, two sets of electronic health records (EHR) tables were used for model training purposes. The EHR records were obtained independently from separate institutions, but contain a large set of common parameters (potential biomarkers - e.g. proteins, RNAs, or metabolites - and routine laboratory hematological testing). For both sets of data, individuals within each set were assigned an International Classification of Diseases, Tenth Revision code (ICD-10) in accordance with a diagnosis (e.g. bacterial: A18.12 tuberculosis of bladder, viral: U07.1 COVID-19) to classify as a type of bacterial or viral infection. The EHR tables from the first set contains 1,628 individuals with an admission diagnosis of bacterial (801) and viral (827) infections. Measurements of an array of parameters were acquired as a first available measurement. The EHR tables from the second set contains a set of 16,239 individuals with a discharge diagnosis as a mean measurement during hospitalization of a bacterial (10,254) and viral (5,985) infections. Measurements of an array of parameters were acquired as a mean measurement during hospitalization.

[0068] For both sets parameters, fourteen different physiological parameters (i.e. absolute EOS count, absolute neutrophil, RBC count, monocyte distribution width, absolute baso count, absolute lymph count, hemoglobin, WBC count, RDW, absolute mono count, hematocrit, MPV, platelet count, and NLR) were identified as containing predictive value in determining bacterial versus viral infections. These features values were used to train a model for predicting bacterial versus viral infections. The first set (1,628 individuals) were analyzed using Logistic regression with nested 10- fold cross validation model. The second set were analyzed using XGBoost with nested 10-fold crossvalidation model. SHAP values for the combined data sets can be visualized in FIGs 6 and 7 for cell counting population parameters and combined cell population / molecular parameters respectively. For the first data set, an AUROC (mean with 95% CI) of 0.879, a sensitivity (mean with 95% CI) of 78.3%, and Specificity (mean with 95% CI) of 81% was obtained. For the second data set, an AUROC (mean with 95% CI) of 0.817, a sensitivity (mean with 95% CI) of 90.2%, and Specificity (mean with 95% CI) of 56.6% was obtained. It is noteworthy that the differences in the sensitivity and specificity may be due to data set that contains a greater number of bacterial diagnosis. Nevertheless, the model can be further refined using alternate types data sets to optimize for the specific applications.

[0069] Turning to FIG. 5, FIG. 5 is an example data flow illustrating an example process 500 for generating a prediction probability for an individual utilizing one or more deployed Al model(s), in accordance with some embodiments of the present disclosure. It should be understood that this and other arrangements described herein are set forth only as examples. Other arrangements and elements (e.g., machines, interfaces, functions, orders, groupings of functions, etc.) may be used in addition to or instead of those shown, and some elements may be omitted altogether. Further, many of the elements described herein are functional entities that may be implemented as discrete or distributed components or in conjunction with other components, and in any suitable combination and location. Various functions described herein as being performed by entities may be carried out by hardware, firmware, and / or software.

[0070] At a high level, the process 500 may include one or more Al model(s) 506 receiving prepared data 504 that is encoded from raw data 502. The raw data 502 may be output from an analyzer (e.g., analyzer 102) or may be stored in an EHR database (e.g., EHR database 106). For example, an analyzer may output cell population parameters (e.g., eosinophil count, neutrophil count, MDW, RBC count, basophil count, lymphocyte count, a measure of hemoglobin, WBC count, RDW, monocyte count, a measure of hematocrit level, a measure of MPV, platelet count, or a combination thereof),ratios thereof (e.g., neutrophil to lymphocyte ratio (NLR), eosinophil to lymphocyte ratio (ELR), lymphocyte to monocyte ratio (LMR), eosinophil-to-monocyte ratio (EMR), platelet-to-lymphocyte ratio (PLR), platelet-to-neutrophil ratio, monocyte to lymphocyte ratio (MLR), mean platelet volume- to-platelet count (MPV / PC) ratio, basophil to lymphocyte (BLR) ratio, and so forth), or any combination thereof. The analyzer may transmit the raw data output to an analysis engine (e.g., analysis engine 300 of FIG. 3) maintaining the one or more deployed Al model(s). For another example, an analysis engine may transmit a request for data from the EHR server 104. The request may include a computer readable request for specific parameters that are stored in the EHR database 106 for the particular patient of interest (e.g., the patient for which the prediction is desired). The specific parameters may include one or more cell population parameters, molecular parameters, patient demographic parameters, or a combination thereof.

[0071] In some embodiments of process 500, the raw data 502 is encoded by one or more encoders to generate prepared data 504. The encoding process may include feature extraction, which refers to the computational identification the most important features of the data and discards the rest. The extracted features are then transformed into a numerical format that can be used as input into the Al model. The encoding process may also involve normalization or scaling of the raw data 502 to ensure that all features have a similar magnitude and range, which may improve the performance of some embodiments of the Al model. Techniques such as mean normalization and min-max scaling can be used to facilitate normalization and / or scaling.

[0072] The prepared data 504 is provided as an input to the one or more layers of the Al model(s) and output data 508 is generated. The output data 508 may include data that is not in a human understandable format. For example, the output data 508 may be vector, an array, probability distribution, hexadecimal representation of a value, or a classification. Accordingly, in some embodiments, the output data 508 is decoded by one or more decoders to generate a predictionprobability of a bacterial or viral infection. Alternatively, the Al model 506 may be configured such that the output 508 is in a human understandable format. For example, the prediction may be a numerical value scaled such that one end of the scale represents the highest probability of a bacterial infection and the opposite end of the scale represents the highest probability of a viral infection. For another example the prediction may be a textual statement of the prediction (e.g., patient is predicted to have a bacterial infection) and a numerical expression of probability (e.g., a percentage). In some embodiments, the output data 508 or the decoded representation of output data 508 is transmitted to a user device (e.g., user device 110 of FIG. 1) for presentation and / or further evaluation 510.

[0073] With general reference to FIG. 6, a bacterial versus viral signature can be established using cell population parameters of a trained machine learning model. Individual feature values (absolute EOS count, absolute neutrophil, RBC count, monocyte distribution width, absolute baso count, absolute lymph count, hemoglobin, WBC count, RDW, absolute mono count, hematocrit, MPV, platelet count, and NLR) of the cell population parameters can assessed for prediction probability by calculating SHapley Additive exPlanations (SHAP) values. For demonstrative purposes, individual feature SHAP values were generated from a machine learning model using patient data from a first institution. The SHAP values, as illustrated in FIG 6, can be used to assess a predictive probability for each of the individual features independently or in combination.

[0074] With specific reference to FIG. 6, each dot in the graph represents a SHAP value for an individual feature value for a patient that has been correctly diagnosed with a bacterial or viral infection. A positive SHAP value adds predictive probability favoring a bacterial infection. A negative SHAP values adds predictive probability favoring a viral infection. In considering FIG. 6, blue and red colored dots are used to indicate a high or low values for an individual feature value. For a lower value of a feature and blue-colored dot is allocated whereas for a higher value of a feature a red- colored dot is allocated. For example, in considering the “Absolute EOS count,” a high density of reddots are located in the positive region of the SHAP plot. Consequently, using the trained model of FIG. 6, a patient with a high Absolute EOS count would be probabilistically correlated with a bacterial infection for that parameter. Conversely, a patient with a low EOS count would be probabilistically correlated with a viral infection for that parameter.

[0075] As previously mentioned, individual feature values of FIG. 6 represent a set of cell population parameters found to have predictive value in determining bacterial versus viral infections. The individual features values are specific measurements from a blood sample of an individual. In particular, an absolute EOS count is a measure of the number of eosinophil cells. An absolute neutrophil count is a measure of the number of neutrophil cells. RBC count is count is a measure of the number of red blood cells. A MDW is the monocyte distribution width or a morphological parameter that is an average volume for monocytes. Absolute basophil count is a measure of the number of basophil cells. Absolute lymph count is a measure of the number of lymphocyte cells. Hemoglobin is a measure of the mass of hemoglobin as a function of volume (i.e. g / dL). WBC is a measure of the number of white blood cells. RDW is the red cell distribution width or a percentage of size deviation of red blood cell when compared to the mean red blood cell size. Absolute mono count is a measure of the number of monocyte cells. Hemocrit is a volume percentage red blood cells as compared the overall blood volume. MPV or mean platelet volume is a measure of the average volume (i.e. fL) of platelets. Platelet count is a measure of the number of platelets. NLR is the neutrophil to lymphocyte ratio or the number of neutrophil cells divided by the number of lymphocyte cells.

[0076] With general reference to FIG. 7, a bacterial versus viral signature can be established using physiological parameters (including cell population parameters and clinical measurements) of a trained machine learning model. Individual feature values (ERS, absolute EOS count, absolute neutrophil, c-reactive protein, WBC UA, RBC count, AST, procalcitonin, MDW, anion gap, ALK phosphatase, RDW, absolute baso count, absolute lymph count, hemoglobin, calcium, absolute monocount, MPV, NLR, hematocrit, platelet count, and O2 sat) of the physiological parameters can be assessed for prediction probability by calculating SHAP values. For demonstrative purposes, individual feature SHAP values were generated from a machine learning model using patient data from the first institution. The SHAP values, as illustrated in FIG 7, can be used to assess a predictive probability, of a viral versus a bacterial infection, for each of the individual features independently or combination.

[0077] With specific reference to FIG. 7, each dot in the graph represents a SHAP value for an individual feature value for a patient that has been correctly diagnosed with a bacterial or viral infection. A positive SHAP value adds predictive probability favoring a bacterial infection. A negative SHAP values adds predictive probability favoring a viral infection. In considering FIG 10, blue and red colored dots are used to indicate a high or low values for an individual feature value. For a lower value of a feature and blue-colored dot is allocated whereas for a higher value of a feature a red- colored dot is allocated. For example, in considering the “Absolute EOS count,” a high density of red dots are located in the positive region of the SHAP plot. Consequently, using the trained model of FIG. 7, a patient with a high Absolute EOS count would be probabilistically correlated with a bacterial infection for that parameter. Conversely, a patient with a low EOS count would be probabilistically correlated with a viral infection for that parameter.

[0078] As previously mentioned, individual feature values of FIG. 7 represent a set of physiological parameters found to have predictive value in determining bacterial versus viral infections. The individual features values can be obtained by using a blood, serum, plasma, urine, or other body fluid sample of an individual to conduct specific measurements. The individual features for the physiological parameters of FIG. 7 include: a erythrocyte sedimentation rate (“ERS”) or measure of rate at which red blood cells settle; an “absolute EOS count” or a measurement of the number of eosinophil cells; an “absolute neutrophil” count or a measurement of the number ofneutrophil cells; “c-reactive protein” or the mass of c-reactive protein as a function of volume (i.e. mg / L); “WBC, UA” or the number of white blood cell in a sample of urine, “RBC count” or a measure or the number of red blood cells; “AST” or a measure of the units of aspartate aminotransferase as a function of volume (U / L); monocyte distribution width (“MDW”) or a morphological parameter that is an average volume for monocytes; “procalcitonin” or a mass of procalcitonin as a function of volume (i.e. ng / mL), “anion gap” or quantity difference between cations and anions as a function of volume (i.e. mmol / L), “alk phosphatase” or a measure of the units of alkaline phosphatase as a function of volume as a function of volume (i.e. U / L); a red cell distribution width (“RDW”) or a percentage of size deviation of red blood cell when compared to the mean red blood cell size; an “absolute lymph count” or a measure of the number of lymphocyte cells; an “absolute baso count” or a measure of the number of basophil cells; “hemoglobin” or a measure of the mass of hemoglobin as a function of volume (i.e. g / dL); “calcium” or the mass of calcium as a function of volume (i.e. mg / mL); an “absolute mono count” or a measure of the number of monocyte cells; a mean platelet volume (“MPV”) or a measure of the average volume (i.e. fL) of platelets; a neutrophil to lymphocyte ratio (“NLR”) or the number of neutrophil cells divided by the number of lymphocyte cells; a “hemocrif ’ or a volume percentage red blood cells as compared the overall blood volume; a “platelet count” or a measure of the number of platelets; an “02 sat” or a measure of percentage of hemoglobin bound to oxygen compared to unbound hemoglobin.

[0079] With general reference to FIG. 8, FIG. 8 depicts a chart 800 providing a comparison of the monocyte distribution width (MDW) can be compared for individuals diagnosed with viral and bacterial infections. For both viral and bacterial diagnosis, it can be shown that that MDW levels are elevated. Viral infections generally result in higher level of elevation than bacterial infections. However, using combinations of markers, parameters, measurements (e.g., cell population parametersand clinical measurements), or a combination thereof offer a greater ability to discriminate clinical phenotypes.

[0080] With specific reference to FIG. 8, distribution data of MDW is represented by a box- and-whisker plot, for viral infection 802 and bacterial infection 804 diagnosis. In considering the viral infection 802, box 806 represents the lower quartile, whereas box 808 represents the upper quartile of MDW values for viral diagnosis. Line 810, where box 806 and box 808 intersect, represents the median MDW values for viral diagnosis. Whisker lines extend to one and half times the interquartile range to establish a boundary for an upper limit of MDW for a viral diagnosis. Points 812 outside of the upper limit are considered outlying MDW values for a viral diagnosis. In considering the bacterial infection 804, box 814 represents the lower quartile, whereas box 816 represents the upper quartile of MDW values for bacterial diagnosis. Line 818, where box 814 and 816 intersect, represents the median MDW values for bacterial diagnosis. Whisker lines extend to one and half times the interquartile range to establish a boundary for an upper limit and a lower limit of MDW for a bacterial diagnosis. Points 820 outside of the upper limit are considered outlying MDW values for a bacterial diagnosis.

[0081] In comparing both the viral plot 802 and the bacterial plot 804, general observations can be made about MDW values for viral versus bacterial infection. The upper quartile 808 and lower quartile 806 are high in value for viral diagnosis in comparison to upper quartile 816 and lower quartile 814 for bacterial infection.

[0082] FIG. 9 illustrates an example method 900 for generating and communicating a prediction probability of one of two classes of infection based on the output of an Al model. The classes include a viral infection and a bacterial infection. Although the example method 900 depicts a particular sequence of operations, the sequence may be altered without departing from the scope of the present disclosure. For example, some of the operations depicted may be performed in parallel orin a different sequence that does not materially affect the function of the method 900. In other examples, different components of an example device or system that implements the method 900 may perform functions at substantially the same time or in a specific sequence.

[0083] According to some examples, the method includes analyzing a blood, serum, plasma, urine, or other body fluid sample drawn from an individual for a hematological cell parameter such as a cell population parameter at step 902. For example, the analyzer 102 illustrated in FIG. 1 may analyze a blood sample drawn from an individual for a cell population parameter. The cell population parameter can be selected from eosinophil count, neutrophil count, monocyte distribution width (MDW), red blood cell (RBC) count, basophil count, lymphocyte count, a measure of hemoglobin, white blood cell (WBC) count, red cell distribution width (RDW), monocyte count, a measure of hematocrit level, a measure of mean platelet volume (MPV), platelet count, ratios thereof, or a combination thereof.

[0084] According to some examples, the method includes determining, for the individual, a clinical measurement or an additional cell population parameter at step 904. For example, the analyzer 102 illustrated in FIG. 1 may determine, for the individual, a clinical measurement or an additional cell population parameter. The clinical measurement can be selected from anion gap, a measure of procalcitonin (PCT), a measure of C-reactive protein (CRP), alkaline phosphatase (ALK), aspartate aminotransferase (AST), calcium, oxygen (O2) saturation, ratios thereof.

[0085] According to some examples, the method includes generating an input data set for the individual including cell population parameters, clinical measurements, and / or molecular parameters at step 906. Step 906 may include receiving output from an analyzer (e.g., analyzer 102). For example, an analyzer may output cell population parameters (e.g., eosinophil count, neutrophil count, MDW, RBC count, basophil count, lymphocyte count, a measure of hemoglobin, WBC count, RDW, monocyte count, a measure of hematocrit level, a measure of MPV, platelet count, or a combinationthereof), ratios thereof (e.g., neutrophil to lymphocyte ratio (NLR), eosinophil to lymphocyte ratio (ELR), lymphocyte to monocyte ratio (LMR), eosinophil-to-monocyte ratio (EMR), platelet-to- lymphocyte ratio (PLR), platelet-to-neutrophil ratio, monocyte to lymphocyte ratio (MLR), mean platelet volume-to-platelet count (MPV / PC) ratio, basophil to lymphocyte (BLR) ratio, and so forth), clinical measurements (e.g., erythrocyte sedimentation rate (ESR), anion gap, a measure of procalcitonin (PCT), a measure of C-reactive protein (CRP), a measure of alkaline phosphatase (ALK), a measure of aspartate aminotransferase (AST), a measure of calcium, or oxygen (O2) saturation), molecular parameters (e.g., protein or mRNA levels of vimentin (VIM), TNF-related apoptosis-inducing ligand (TRAIL), procalcitonin (PCT), C Reactive Protein (CRP), interferon y- induced protein (IP- 10), Myxovirus resistance A (MxA), Cluster of Differentiation 64 (CD64), human neutrophile lipocalin (HNL), blood urea nitrogen (BUN), lactate, heparin-binding protein (HBP), adrenomedullin (ProADM), bioavailable adrenomedullin (bio-ARM), midregional adrenomedullin (MR-proADM), interleukin 6 (IL-6), ABL Proto-Oncogene 1 (ABL1), Interferon Regulatory Factor 9 (IRF9), Integrin Subunit Alpha M (ITGAM), Lymphocyte Antigen 6 Family Member E (LY6E), Proline-Serine-Threonine Phosphatase Interacting Protein 2 (PSTPIP2), runt-related transcription factor 1 (RUNX1), and so forth), or any combination thereof. The analyzer may transmit the raw data output to an analysis engine (e.g., analysis engine 300 of FIG. 3) maintaining the one or more deployed Al model(s).

[0086] Additionally, or alternatively, step 906 may include collecting data stored in an EHR database (e.g., EHR database 106). For example, an analysis engine may transmit a request for data from the EHR server 104. The request may include a computer readable request for specific parameters that are stored in the EHR database 106 for the particular patient of interest (e.g., the patient for which the prediction is desired). The specific parameters may include one or morehematological cell parameters (e.g., cell population parameters), clinical measurements, molecular parameters, patient demographic parameters, or a combination thereof.

[0087] According to some examples, the method includes providing the input data set as input to a trained model configured to generate an output value at step 908. For example, the model execution environment 108 illustrated in FIG. 1 may provide the input data set as input to a trained model maintained by an analysis engine (e.g. analysis engine 300 of FIG. 3) configured to generate an output value.

[0088] According to some examples, the method includes responsive to the output value of the trained model exceeding a predetermined threshold, communicating an indication of a prediction probability favoring a bacterial infection of the individual at step 910. For example, the EHR database 106 illustrated in FIG. 1 may communicate an indication to a user device 110. The indication may include a prediction that the patient associated with Al model’s analysis has a bacterial infection. For another example, the indication may include a prediction that the patient associated with Al model’s analysis has a bacterial infection. Responsive to the output value of the trained model exceed a predetermined threshold, communicating an indication of a prediction probability favoring a bacterial infection of the individual. Additionally, the communication can be configured to occur based on a triggering event. For example, in some aspects of step 910, the communication occurs in response to a care provider’s utilization of user device 110 to access the results of hematological or metabolic testing of the patient.

[0089] According to some examples, the method includes responsive to the output value of the trained model exceeding a predetermined threshold, communicating a treatment recommendation, an additional diagnostic test recommendation, a triage recommendation, a prognosis, or a medical prescription. In other aspects, the results can be used to generate a recommendation for an additional diagnostic test that can further distinguish between a bacterial or a viral infection. The additionaldiagnostic test to distinguish between bacterial infection or viral infection may be selected from one or more of, PCR, RT-PCR, qPCR, ELISA, immunoassay, flow cytometry, or lateral flow assay. In aspects, the additional diagnostic test to distinguish between bacterial infection or viral infection comprises detecting host-derived protein biomarkers in response to a viral or bacterial infection. In aspects, the additional diagnostic test to distinguish between bacterial infection or viral infection comprises detecting one or both of a host gene expression response to a viral or bacterial infection. In aspects, the additional diagnostic test to distinguish between bacterial infection or viral infection comprises detecting one or more indicators of infection selected from vimentin (VIM), TNF-related apoptosis-inducing ligand (TRAIL), procalcitonin (PCT), C Reactive Protein (CRP), interferon y- induced protein (IP- 10), Myxovirus resistance A (MxA), Cluster of Differentiation 64 (CD64), human neutrophile lipocalin (HNL), blood urea nitrogen (BUN), lactate, heparin-binding protein (HBP), pro- adrenomedullin (ProADM), interleukin 6 (IL-6), ABL Proto-Oncogene 1 (ABL1), Interferon Regulatory Factor 9 (IRF9), Integrin Subunit Alpha M (ITGAM), Lymphocyte Antigen 6 Family Member E (LY6E), Proline-Serine-Threonine Phosphatase Interacting Protein 2 (PSTPIP2), and runt- related transcription factor 1 (RUNX1). In aspects, the additional diagnostic test to distinguish between bacterial infection or viral infection comprises determining expression levels of RUNX1, ITGAM, PSTPIP2, LY6E, and IRF9. The additional diagnostic tool may comprise detection of a host gene expression response that discriminates between a bacterial infection and a viral infection based on detecting the expression level of a combination of ABL1, IRF9, ITGAM, LY6E, PSTPIP2 and RUNX1 in biological samples from the human subject and determining whether the human subject has a bacterial or viral infection based on those expression levels, as described in WO2022197351. In brief, the detection of a host gene expression response may comprise: determining if a subject has a bacterial infection, a viral infection, or a non-infectious cause of fever, further comprising: a) determining the expression levels of RUNX1, ITGAM, PSTPIP2, LY6E, IRF9 and ABL1 in abiological sample from the subject; b) determining that the expression level of ABL1 is greater than a first predetermined cutoff value; c) determining a viral infection score and a bacterial infection score based on the expression levels of RUNX1, ITGAM, PSTPIP2, LY6E and IRF9; d) comparing the viral infection score to a second predetermined cutoff value and the bacterial infection score to a third predetermined cutoff value; and e) determining: that the subject has a viral infection when the viral infection score is greater than the second predetermined cutoff value or that the subject does not have viral infection when the viral infection score is less than the second predetermined cutoff value, that the subject has a bacterial infection when the bacterial infection score is greater than the third predetermined cutoff value or the subject does not have a bacterial infection when the bacterial infection score is less than or equal to the third predetermined cutoff value, and that the subject has a non-infectious cause of fever in the subject when the viral infection score is less than the second predetermined cutoff value and the bacterial infection score is less than the third predetermined cutoff value. In further aspects, the additional diagnostic test may be a single biomarker test, such as CRP or PCT. In further aspects, the additional diagnostic test may employ a a protein panel, for example, PCT+IP-10+CRP or CRP + MxA. In further aspects, the additional diagnostic test may be a gene expression panel, such as a 10-Gene signature panel (as described in Bhattacharya et al, Sci Rep. 2017; 7:6548) or 130-Gene signature panel (as described in Tsalik et al, Sci Transl Med. 2016; 8: 322ral 1). In further aspects, the additional diagnostic test may be a test administered via a wearable technology, for example an ADAMM-RSM or Fitbit.

[0090] FIG. 10 illustrates another example method 1000 for generating and communicating a prediction probability of one of two classes of infection based on the output of an Al model. The classes include a viral infection and a bacterial infection. Although the example method 1000 depicts a particular sequence of operations, the sequence may be altered without departing from the scope ofthe present disclosure. For example, some of the operations depicted may be performed in parallel or in a different sequence that does not materially affect the function of the method 1000. In other examples, different components of an example device or system that implements the method 1000 may perform functions at saubstantially the same time or in a specific sequence.

[0091] According to some examples, method 1000 includes analyzing a blood sample drawn from an individual for a monocyte distribution width (MDW) value for the individual at step 1002. For example, the analyzer 102 illustrated in FIG. 1 may analyze a blood sample drawn from an individual for a monocyte distribution width (MDW) value for the individual. According to some examples, method 1000 includes determining, for the individual, an eosinophil count and neutrophil count at step 1004. For example, the analyzer 102 illustrated in FIG. 1 may determine, for the individual, an eosinophil count and neutrophil count.

[0092] According to some examples, method 1000 includes generating an input data set for the individual including the MDW value, the eosinophil count, and the neutrophil count at block 1006. For example, the analysis engine 300 illustrated in FIG. 3 may generate an input data set for the individual including the MDW value, the eosinophil count, and the neutrophil count.

[0093] According to some examples, method 1000 includes providing the input data set as input to a trained model configured to generate an output value at block 1008. For example, the analysis engine 300 illustrated in FIG. 3 may provide the input data set as input to a trained model configured to generate an output value.

[0094] According to some examples, the method includes communicating an indication of a prediction probability favoring a bacterial infection of the individual responsive to the output value of the trained model exceed a predetermined threshold, at block 1010. For example, the user device 110 illustrated in FIG. 1 may responsive to the output value of the trained model exceed a predeterminedthreshold, communicating an indication of a prediction probability favoring a bacterial infection of the individual.

[0095] In the preceding description, for the purposes of explanation, numerous details have been set forth in order to provide an understanding of various embodiments of the present technology. It will be apparent to one skilled in the art, however, that certain embodiments may be practiced without some of these details, or with additional details, or in varied combinations or subcombinations of features of the embodiments.

[0096] Having described several embodiments, it will be recognized by those of skill in the art that various modifications, alternative constructions, and equivalents may be used without departing from the spirit of the invention. Additionally, a number of well-known processes and elements have not been described in order to avoid unnecessarily obscuring the present invention. Additionally, details of any specific embodiment may not always be present in variations of that embodiment or may be added to other embodiments.

[0097] Where a range of values is provided, it is understood that each intervening value, to the tenth of the unit of the lower limit unless the context clearly dictates otherwise, between the upper and lower limits of that range is also specifically disclosed. Each smaller range between any stated value or intervening value in a stated range and any other stated or intervening value in that stated range is encompassed. The upper and lower limits of these smaller ranges may independently be included or excluded in the range, and each range where either, neither, or both limits are included in the smaller ranges is also encompassed within the invention, subject to any specifically excluded limit in the stated range. Where the stated range includes one or both of the limits, ranges excluding either or both of those included limits are also included.

[0098] As used herein and in the appended claims, the singular forms “a”, “an”, and “the” include plural referents unless the context clearly dictates otherwise. Thus, for example, reference to“a method” includes a plurality of such methods and reference to “the transducer” includes reference to one or more transducers and equivalents thereof known to those skilled in the art, and so forth. The invention has now been described in detail for the purposes of clarity and understanding. However, it will be appreciated that certain changes and modifications may be practice within the scope of the appended claims.

[0099] As used herein, “parameter” may refer to a property measured from a blood, serum, plasma, urine, or other body fluid sample (e.g., cell population data parameters) or independent variables in a function. The broader definition of “parameter” including both types apply unless context makes clear otherwise. A “characteristic” refers to a property of a blood, serum, plasma, urine, or other body fluid sample and not an independent variable in a function.

[0100] As used herein and in the appended claims, a “reflex test” refers to a test that is performed because certain data or circumstances indicate that additional testing would be helpful. Reflex testing often occurs because of results which could have significant clinical effects. In such cases, the reflex testing may be intended to verify clinically important test result. In the context of this disclosure, reflex testing may occur to confirm a result, to increase the confidence in a result (e.g., by extended sampling), and / or because the result of one or more routine or specially ordered tests may indicate that another panel of tests, such as a panel of tests for a specific condition, is indicated. The reflex testing may occur because a medical practitioner ordered a condition-specific panel, and a set of one or more primary tests indicated that additional testing was advisable (e.g., one or more results was consistent with a diagnosis of the condition). The reflex testing for a condition-specific panel may occur even if the medical practitioner did not order a condition-specific panel, if the results indicate that the condition-specific panel should be tested. Alternately, the testing described herein as reflex testing may be part of the routine processing of a condition-specific panel (e.g., not actually reflextesting, but rather routine testing vis-a-vis an order for a condition-specific panel, regardless of the results of whatever tests happen to be run first in time).ILLUSTRATIVE COMBINATIONS

[0101] The following examples relate to various non-exhaustive ways in which the teachings herein may be combined or applied. It should be understood that the following examples are not intended to restrict the coverage of any claims that may be presented at any time in this application or in subsequent filings of this application. No disclaimer is intended. The following examples are being provided for nothing more than merely illustrative purposes. It is contemplated that the various teachings herein may be arranged and applied in numerous other ways. It is also contemplated that some variations may omit certain features referred to in the below examples. Therefore, none of the aspects or features referred to below should be deemed critical unless otherwise explicitly indicated as such at a later date by the inventors or by a successor in interest to the inventors. If any claims are presented in this application or in subsequent filings related to this application that include additional features beyond those referred to below, those additional features shall not be presumed to have been added for any reason relating to patentability.

[0102] Example 1

[0103] A method of detecting and differentiating bacterial and viral infections comprising: acquiring a hematological cell parameter for a sample from an individual; obtaining, a set of clinical measurements for an organism associated with the sample which includes at least one of an erythrocyte sedimentation rate (ESR), anion gap, a measure of procalcitonin (PCT), a measure of C- reactive protein (CRP), a measure of alkaline phosphatase (ALK), a measure of aspartate aminotransferase (AST), a measure of calcium, or oxygen (O2) saturation; generating an input data set for the organism including the hematological cell parameter and the clinical measurements;providing the input data set as input to a trained model configured to generate an output value; comparing the output value of the trained model to a predetermined threshold value to determine a probability associated with a bacterial infection or a viral infection.

[0104] Example 2

[0105] The method of example 1, wherein the sample is obtained from blood, serum, plasma, urine, or other body fluid drawn from a human.

[0106] Example 3

[0107] The method of example 1 or 2, wherein when an output value exceeds the predetermined threshold value, the method further comprises modifying a report to include an indication of suspected bacterial infection.

[0108] Example 4

[0109] The method of any one of examples 1-3, wherein when an output value is less than or equal to a predetermined threshold value, the method further comprises modifying a report to include an indication of suspected viral infection.

[0110] Example 5

[0111] The method of any one of examples 1-4, further comprising: communicatively coupling with a database; and retrieving, from the database values associated with the hematological cell parameter, the erythrocyte sedimentation rate (ESR), the anion gap, the measure of PCT, the measure of CRP, the measure of ALK, the measure of AST, the measure of calcium, or the O2 saturation for the organism.

[0112] Example 6

[0113] The method of example 1 or 2, wherein the trained model is trained using a log loss function to optimize true positives for a bacterial.

[0114] Example 7

[0115] The method of any one of examples 1-6, wherein the hematological cell parameter is selected from a cell population parameter, complete blood count (CBC), white blood cell count, mean platelet volume (MPV), ratios thereof, or a combination thereof.

[0116] Example s

[0117] The method of example 7, wherein the hematological cell parameter is a cell population parameter.

[0118] Example 9

[0119] The method of example 7 or 8, wherein the cell population parameter is selected from a monocyte cell population parameter, a granulocyte cell population parameter, a lymphocyte cell population parameter, a red blood cell population parameter, white blood cell count, platelet count, mean platelet volume (MPV), ratios thereof, or a combination thereof.

[0120] Example 10

[0121] The method of example 8 or 9, wherein the cell population parameter comprises a monocyte cell population parameter.

[0122] Example 11

[0123] The method of example 10, wherein the monocyte cell population parameter is selected from monocyte distribution width (MDW), monocyte count, standard deviation in monocyte volume, mean monocyte volume, reactive monocytes, ratios thereof, or a combination thereof, preferably wherein the monocyte cell population parameter is MDW.

[0124] Example 12

[0125] The method of any one of examples 8-11, wherein cell population parameter comprises a granulocyte cell population parameter.

[0126] Example 13

[0127] The method of example 12, wherein the granulocyte cell population parameter comprises eosinophil count, neutrophil count, basophil count, neutrophil volume distribution width, standard deviation in the volume of neutrophils, ratios thereof, or a combination thereof

[0128] Example 14

[0129] The method of any one of examples 8-13, wherein cell population parameter comprises a red blood cell population parameter.

[0130] Example 15

[0131] The method of example 14, wherein the red blood cell population parameter comprises red blood cell count, red cell distribution width (RDW), a measure of hemoglobin, a measure of hematocrit level, ratios thereof, or a combination thereof.

[0132] Example 16

[0133] The method of any one of examples 8-15, wherein cell population parameter comprises a lymphocyte cell population parameter.

[0134] Example 17

[0135] The method of example 16, wherein the lymphocyte cell population parameter comprises lymphocyte count, standard deviation in lymphocyte volume, mean lymphocyte volume, reactive lymphocyte, ratios thereof, or a combination thereof.

[0136] Example 18

[0137] The method of any one of examples 8-17, wherein cell population parameter comprises a neutrophil to lymphocyte ratio (NLR), eosinophil to lymphocyte ratio (ELR), lymphocyte to monocyte ratio (LMR), eosinophil-to-monocyte ratio (EMR), platelet-to-lymphocyte ratio (PLR), platelet-to-neutrophil ratio, monocyte to lymphocyte ratio (MLR), mean platelet volume-to-platelet count (MPV / PC) ratio, basophil to lymphocyte (BLR) ratio, or combinations thereof.

[0138] Example 19

[0139] The method of example 18, wherein the cell population parameter comprises a neutrophil to lymphocyte ratio (NLR).

[0140] Example 20

[0141] The method of any one of examples 7-19, wherein the hematological cell parameter comprises complete blood count (CBC).

[0142] Example 21

[0143] The method of any one of example 7-20, wherein the hematological cell parameter comprises white blood cell count.

[0144] Example 22

[0145] The method of any one of examples 7-21, wherein the hematological cell parameter comprises mean platelet volume (MPV).

[0146] Example 23

[0147] The method of any one of examples 1-22, wherein the clinical measurement includes erythrocyte sedimentation rate (ESR), a measure of procalcitonin (PCT), a measure of C-reactive protein (CRP), or a combination thereof.

[0148] Example 24

[0149] The method of any one of examples 1-23, wherein the input data set for the individual includes erythrocyte sedimentation rate (ESR), eosinophil count, neutrophil count, a measure of C- reactive protein (CRP), white blood cell (WBC) count, red blood cell (RBC) count, a measure of aspartate aminotransferase (AST), a measure of procalcitonin (PCT), monocyte distribution width (MDW), anion gap, a measure of alkaline phosphatase (ALK), red cell distribution width (RDW), basophil count, lymphocyte count, a measure of hemoglobin, calcium, monocyte count, a measure of mean platelet volume (MPV), neutrophil to lymphocyte ratio, a measure of hematocrit level, platelet count, oxygen (O2) saturation, ratios thereof, or combinations thereof.

[0150] Example 25

[0151] The method of any one of examples 1-24, wherein the input data set for the individual includes erythrocyte sedimentation rate (ESR), eosinophil count, neutrophil count, a measure of C- reactive protein (CRP), white blood cell (WBC) count, red blood cell (RBC) count, a measure of aspartate aminotransferase (AST), a measure of procalcitonin (PCT), monocyte distribution width (MDW), anion gap, a measure of alkaline phosphatase (ALK), red cell distribution width (RDW), basophil count, lymphocyte count, a measure of hemoglobin, calcium, monocyte count, a measure of mean platelet volume (MPV), neutrophil to lymphocyte ratio, a measure of hematocrit level, platelet count, and oxygen (O2) saturation.

[0152] Example 26

[0153] The method of any one of examples 1-25, wherein the input data set for the individual additionally includes an eosinophil count, neutrophil count, or a combination thereof.

[0154] Example 27

[0155] The method of any one of example 7-26, wherein the input data set for the individual additionally includes at least one of red blood cell (RBC) count, basophil count, lymphocyte count, a measure of hemoglobin, white blood cell (WBC) count, red cell distribution width (RDW), monocyte count, a measure of mean platelet volume (MPV), platelet count, or neutrophil to lymphocyte ratio.

[0156] Example 28

[0157] The method of any one of examples 1-26, wherein the input data set for the individual additionally includes at least one of red blood cell (RBC) count, basophil count, lymphocyte count, monocyte count, platelet count, or neutrophil to lymphocyte ratio.

[0158] Example 29

[0159] The method of any one of examples 1-28, wherein volume conductivity scatter parameters of monocytes are measured for a portion of the sample to generate MDW measurements.

[0160] Example 30

[0161] The method of any one of examples 1-28, wherein light scatter parameters of monocytes, heme flow imaging, or a combination thereof, are measured for a portion of the sample to generate hematological cell parameter (e.g., MDW) measurements.

[0162] Example 31

[0163] The method of any one of examples 1-30, wherein colorimetric parameters, turbidimetric parameters, latex agglutination, ion selective electrode (indirect potentiometry), or a combination thereof, are measured for a portion of the sample to generate clinical measurements.

[0164] Example 32

[0165] The method of any one of examples 1-31, wherein the clinical measurements are measured for a portion of the sample to generate diagnostic information.

[0166] Example 33

[0167] The method of any one of examples 1-32, wherein the set of clinical measurements and hematological cell parameters are measured using a common instrument.

[0168] Example 34

[0169] The method of any one of examples 1-33, wherein the probability associated with a bacterial infection or a viral infection are used to generate a treatment recommendation, an additional diagnostic test recommendation, a triage recommendation, a prognosis, or a medical prescription.

[0170] Example 35

[0171] The method of any one of example 1-34, wherein the probability associated with a bacterial infection or a viral infection are used to order an additional diagnostic test to further distinguish between bacterial infection or viral infection.

[0172] Example 36

[0173] The method of example 25, wherein the additional diagnostic test to distinguish between bacterial infection or viral infection is selected from one or more of, PCR, RT-PCR, qPCR, ELISA, immunoassay, flow cytometry, or lateral flow assay.

[0174] Example 37

[0175] The method of example 25, wherein the additional diagnostic test to distinguish between bacterial infection or viral infection comprises detecting host-derived protein biomarkers in response to a viral or bacterial infection.

[0176] Example 38

[0177] The method of example 25, wherein the additional diagnostic test to distinguish between bacterial infection or viral infection comprises detecting one or both of a host gene expression response to a viral or bacterial infection.

[0178] Example 39

[0179] The method of any one of examples 25-28, wherein the additional diagnostic test to distinguish between bacterial infection or viral infection comprises detecting one or more indicators of infection selected from vimentin (VIM), TNF-related apoptosis-inducing ligand (TRAIL), procalcitonin (PCT), C Reactive Protein (CRP), interferon y-induced protein (IP- 10), Myxovirus resistance A (MxA), Cluster of Differentiation 64 (CD64), human neutrophile lipocalin (HNL), blood urea nitrogen (BUN), lactate, heparin-binding protein (HBP), pro-adrenomedullin (ProADM), interleukin 6 (IL-6), ABL Proto-Oncogene 1 (ABL1), Interferon Regulatory Factor 9 (IRF9), Integrin Subunit Alpha M (ITGAM), Lymphocyte Antigen 6 Family Member E (LY6E), Proline-Serine- Threonine Phosphatase Interacting Protein 2 (PSTPIP2), and runt-related transcription factor 1 (RUNX1).

[0180] Example 40

[0181] The method of example 28 or 29, wherein the additional diagnostic test to distinguish between bacterial infection or viral infection comprises determining expression levels of RUNX1,ITGAM, PSTPIP2, LY6E, and IRF9.

[0182] Example 41

[0183] The method of any one of examples 1-40, wherein the probabilities associated with a bacterial infection or a viral infection are used to rectify an erroneous treatment recommendation or medical prescription.

[0184] Example 42

[0185] Non-transitory computer storage media storing instructions that when executed by one or more processors cause the one or more processors to perform a process comprising: acquiring hematological cell parameter for a sample from an individual; obtaining, clinical measurements for the sample, wherein the clinical measurements comprise: an anion gap, a measure of procalcitonin (PCT), a measure of C-reactive protein (CRP), alkaline phosphatase (ALK), aspartate aminotransferase (AST), calcium, or oxygen (O2) saturation, or a combination thereof; generating an input data set for the individual including the MDW value and the clinical measurements; providing the input data set as input to a trained model configured to generate an output value; comparing the output value of the trained model to predetermined threshold values to determine probabilities associated with a bacterial infection, a viral infection, or a combination or absence thereof.

[0186] Example 43

[0187] The computer storage media of example 42, wherein the sample is obtained from blood, serum, plasma, urine, or other body fluid drawn from the individual.

[0188] Example 44

[0189] The computer storage media of example 43, wherein the individual is human.

[0190] Example 45

[0191] The computer storage media of any one of examples 42-44, wherein the hematological cell parameter (e.g., MDW), the anion gap, the measure of PCT, the measure of CRP, the measure of ALK, the measure of AST, the measure of calcium, or the O2 saturation, or a combination thereof are determined from the sample.

[0192] Example 46

[0193] The computer storage media of any one of examples 42-45, further comprising: communicatively coupling with a database; and retrieving, from the database values associated with the hematological cell parameters (e.g., MDW), the anion gap, the measure of PCT, the measure of CRP, the measure of ALK, the measure of AST, the measure of calcium, or the O2 saturation.

[0194] Example 47

[0195] The computer storage media of any one of examples 42-46, wherein the trained model is trained using a log loss function to optimize true positives for bacterial or viral infection.

[0196] Example 48

[0197] The computer storage media of any one of examples 42-47, wherein the input data set for the individual additionally includes an eosinophil count, neutrophil count, and erythrocyte sedimentation rate (ESR) or a combination thereof.

[0198] Example 49

[0199] The computer storage media of any one of examples 42-48, wherein the input data set for the individual additionally includes at least one of red blood cell (RBC) count, basophil count, lymphocyte count, a measure of hemoglobin, white blood cell (WBC) count, red cell distribution width (RDW), monocyte count, a measure of mean platelet volume (MPV), platelet count, or neutrophil to lymphocyte ratio.

[0200] Example 50

[0201] The computer storage media of any one of examples 42-49, wherein volume conductivity scatter parameters of monocytes are measured for a portion of the sample to generate diagnostic information.

[0202] Example 51

[0203] The computer storage media of any one of examples 42-49, wherein light scatter parameters of monocytes are measured for a portion of the sample to generate diagnostic information.

[0204] Example 52

[0205] A system comprising: a light scatter detection sensor; one or more processors; non- transitory computer storage media communicating coupled to the one or more processors, the computer storage media storing instructions that when executed by the one or more processors cause the one or more processors to perform a process comprising: generating a hematological cell parameter for an individual based on measurements for a sample; obtaining, clinical measurements for the sample, wherein the clinical measurements comprise: an anion gap, a measure of procalcitonin (PCT), a measure of C-reactive protein (CRP), alkaline phosphatase (ALK), aspartate aminotransferase (AST), calcium, or oxygen (O2) saturation, or a combination thereof; generating an input data set for the individual including the MDW value and the clinical measurements; providing the input data set as input to a trained model configured to generate an output value; comparing the output value of the trained model to predetermined threshold values to determine probabilities associated with a bacterial infection, a viral infection, or a combination or absence thereof.

Claims

CLAIMSWhat is claimed is:

1. A method of detecting and differentiating bacterial and viral infections comprising: acquiring a hematological cell parameter for a sample from an individual; obtaining, a set of clinical measurements for an organism associated with the sample which includes at least one of an erythrocyte sedimentation rate (ESR), anion gap, a measure of procalcitonin (PCT), a measure of C-reactive protein (CRP), a measure of alkaline phosphatase (ALK), a measure of aspartate aminotransferase (AST), a measure of calcium, or oxygen (O2) saturation; generating an input data set for the organism including the hematological cell parameter and the clinical measurements; providing the input data set as input to a trained model configured to generate an output value; comparing the output value of the trained model to a predetermined threshold value to determine a probability associated with a bacterial infection or a viral infection.

2. The method of claim 1, wherein the sample is obtained from blood, serum, plasma, urine, or other body fluid drawn from a human.

3. The method of claim 1 or 2, wherein when an output value exceeds the predetermined threshold value, the method further comprises modifying a report to include an indication of suspected bacterial infection.

4. The method of any one of claims 1-3, wherein when an output value is less than or equal to a predetermined threshold value, the method further comprises modifying a report to include an indication of suspected viral infection.

5. The method of any one of claims 1-4, further comprising: communicatively coupling with a database; and retrieving, from the database values associated with the hematological cell parameter, the erythrocyte sedimentation rate (ESR), the anion gap, the measure of PCT, the measure of CRP, the measure of ALK, the measure of AST, the measure of calcium, or the O2 saturation for the organism.

6. The method of claim 1 or 2, wherein the trained model is trained using a log loss function to optimize true positives for a bacterial.

7. The method of any one of claims 1-6, wherein the hematological cell parameter is selected from a cell population parameter, complete blood count (CBC), white blood cell count, mean platelet volume (MPV), ratios thereof, or a combination thereof.

8. The method of claim 7, wherein the hematological cell parameter is a cell population parameter.

9. The method of claim 7 or 8, wherein the cell population parameter is selected from a monocyte cell population parameter, a granulocyte cell population parameter, a lymphocyte cell population parameter, a red blood cell population parameter, white blood cell count, platelet count, mean platelet volume (MPV), ratios thereof, or a combination thereof.

10. The method of claim 8 or 9, wherein the cell population parameter comprises a monocyte cell population parameter.

11. The method of claim 10, wherein the monocyte cell population parameter is selected from monocyte distribution width (MDW), monocyte count, standard deviation in monocyte volume, mean monocyte volume, reactive monocytes, ratios thereof, or a combination thereof, preferably wherein the monocyte cell population parameter is MDW.

12. The method of any one of claims 8-11, wherein cell population parameter comprises a granulocyte cell population parameter.

13. The method of claim 12, wherein the granulocyte cell population parameter comprises eosinophil count, neutrophil count, basophil count, neutrophil volume distribution width, standard deviation in the volume of neutrophils, ratios thereof, or a combination thereof.

14. The method of any one of claims 8-13, wherein cell population parameter comprises a red blood cell population parameter.

15. The method of claim 14, wherein the red blood cell population parameter comprises red blood cell count, red cell distribution width (RDW), a measure of hemoglobin, a measure of hematocrit level, ratios thereof, or a combination thereof.

16. The method of any one of claims 8-15, wherein cell population parameter comprises a lymphocyte cell population parameter.

17. The method of claim 16, wherein the lymphocyte cell population parameter comprises lymphocyte count, standard deviation in lymphocyte volume, mean lymphocyte volume, reactive lymphocyte, ratios thereof, or a combination thereof.

18. The method of any one of claims 8-17, wherein cell population parameter comprises a neutrophil to lymphocyte ratio (NLR), eosinophil to lymphocyte ratio (ELR), lymphocyte to monocyte ratio (LMR), eosinophil-to-monocyte ratio (EMR), platelet-to-lymphocyte ratio (PLR), platelet-to-neutrophil ratio, monocyte to lymphocyte ratio (MLR), mean platelet volume-to- platelet count (MPV / PC) ratio, basophil to lymphocyte (BLR) ratio, or combinations thereof.

19. The method of claim 18, wherein the cell population parameter comprises a neutrophil to lymphocyte ratio (NLR).

20. The method of any one of claims 7-19, wherein the hematological cell parameter comprises complete blood count (CBC).

21. The method of any one of claim 7-20, wherein the hematological cell parameter comprises white blood cell count.

22. The method of any one of claims 7-21, wherein the hematological cell parameter comprises mean platelet volume (MPV).

23. The method of any one of claims 1-22, wherein the clinical measurement includes erythrocyte sedimentation rate (ESR), a measure of procalcitonin (PCT), a measure of C-reactive protein (CRP), or a combination thereof.

24. The method of any one of claims 1-23, wherein the input data set for the individual includes erythrocyte sedimentation rate (ESR), eosinophil count, neutrophil count, a measure of C-reactive protein (CRP), white blood cell (WBC) count, red blood cell (RBC) count, a measure of aspartate aminotransferase (AST), a measure of procalcitonin (PCT), monocyte distribution width (MDW), anion gap, a measure of alkaline phosphatase (ALK), red cell distribution width (RDW), basophil count, lymphocyte count, a measure of hemoglobin, calcium, monocyte count, a measure of mean platelet volume (MPV), neutrophil to lymphocyte ratio, a measure of hematocrit level, platelet count, oxygen (O2) saturation, ratios thereof, or combinations thereof.

25. The method of any one of claims 1-24, wherein the input data set for the individual includes erythrocyte sedimentation rate (ESR), eosinophil count, neutrophil count, a measure of C-reactive protein (CRP), white blood cell (WBC) count, red blood cell (RBC) count, a measure of aspartate aminotransferase (AST), a measure of procalcitonin (PCT), monocyte distribution width (MDW), anion gap, a measure of alkaline phosphatase (ALK), red cell distribution width (RDW), basophil count, lymphocyte count, a measure of hemoglobin, calcium, monocyte count, a measure of mean platelet volume (MPV), neutrophil to lymphocyte ratio, a measure of hematocrit level, platelet count, and oxygen (O2) saturation.

26. The method of any one of claims 1-25, wherein the input data set for the individual additionally includes an eosinophil count, neutrophil count, or a combination thereof.

27. The method of any one of claim 7-26, wherein the input data set for the individual additionally includes at least one of red blood cell (RBC) count, basophil count, lymphocyte count, a measure of hemoglobin, white blood cell (WBC) count, red cell distribution width (RDW), monocyte count, a measure of mean platelet volume (MPV), platelet count, or neutrophil to lymphocyte ratio.

28. The method of any one of claims 1-26, wherein the input data set for the individual additionally includes at least one of red blood cell (RBC) count, basophil count, lymphocyte count, monocyte count, platelet count, or neutrophil to lymphocyte ratio.

29. The method of any one of claims 1-28, wherein volume conductivity scatter parameters of monocytes are measured for a portion of the sample to generate MDW measurements.

30. The method of any one of claims 1-28, wherein light scatter parameters of monocytes, heme flow imaging, or a combination thereof, are measured for a portion of the sample to generate hematological cell parameter (e.g., MDW) measurements.

31. The method of any one of claims 1-30, wherein colorimetric parameters, turbidimetric parameters, latex agglutination, ion selective electrode (indirect potentiometry), or a combination thereof, are measured for a portion of the sample to generate clinical measurements.

32. The method of any one of claims 1-31, wherein the clinical measurements are measured for a portion of the sample to generate diagnostic information.

33. The method of any one of claims 1-32, wherein the set of clinical measurements and hematological cell parameters are measured using a common instrument.

34. The method of any one of claims 1-33, wherein the probability associated with a bacterial infection or a viral infection are used to generate a treatment recommendation, an additional diagnostic test recommendation, a triage recommendation, a prognosis, or a medical prescription.

35. The method of any one of claim 1-34, wherein the probability associated with a bacterial infection or a viral infection are used to order an additional diagnostic test to further distinguish between bacterial infection or viral infection.

36. The method of claim 25, wherein the additional diagnostic test to distinguish between bacterial infection or viral infection is selected from one or more of, PCR, RT-PCR, qPCR, ELISA, immunoassay, flow cytometry, or lateral flow assay.

37. The method of claim 25, wherein the additional diagnostic test to distinguish between bacterial infection or viral infection comprises detecting host-derived protein biomarkers in response to a viral or bacterial infection.

38. The method of claim 25, wherein the additional diagnostic test to distinguish between bacterial infection or viral infection comprises detecting one or both of a host gene expression response to a viral or bacterial infection.

39. The method of any one of claims 25-28, wherein the additional diagnostic test to distinguish between bacterial infection or viral infection comprises detecting one or more indicators of infection selected from vimentin (VIM), TNF-related apoptosis-inducing ligand (TRAIL), procalcitonin (PCT), C Reactive Protein (CRP), interferon y -induced protein (IP- 10), Myxovirus resistance A (MxA), Cluster of Differentiation 64 (CD64), human neutrophile lipocalin (HNL), blood urea nitrogen (BUN), lactate, heparin-binding protein (HBP), pro-adrenomedullin (ProADM), interleukin 6 (IL-6), ABL Proto-Oncogene 1 (ABL1), Interferon Regulatory Factor 9 (IRF9), Integrin Subunit Alpha M (ITGAM), Lymphocyte Antigen 6 Family Member E (LY6E), Proline- Serine-Threonine Phosphatase Interacting Protein 2 (PSTPIP2), and runt-related transcription factor 1 (RUNX1).

40. The method of claim 28 or 29, wherein the additional diagnostic test to distinguish between bacterial infection or viral infection comprises determining expression levels of RUNX1, ITGAM, PSTPIP2, LY6E, and IRF9.

41. The method of any one of claims 1-40, wherein the probabilities associated with a bacterial infection or a viral infection are used to rectify an erroneous treatment recommendation or medical prescription.

42. Non-transitory computer storage media storing instructions that when executed by one or more processors cause the one or more processors to perform a process comprising: acquiring hematological cell parameter for a sample from an individual; obtaining, clinical measurements for the sample, wherein the clinical measurements comprise: an anion gap, a measure of procalcitonin (PCT), a measure of C-reactive protein(CRP), alkaline phosphatase (ALK), aspartate aminotransferase (AST), calcium, or oxygen (O2) saturation, or a combination thereof; generating an input data set for the individual including the MDW value and the clinical measurements; providing the input data set as input to a trained model configured to generate an output value; comparing the output value of the trained model to predetermined threshold values to determine probabilities associated with abacterial infection, a viral infection, or a combination or absence thereof.

43. The computer storage media of claim 42, wherein the sample is obtained from blood, serum, plasma, urine, or other body fluid drawn from the individual.

44. The computer storage media of claim 43, wherein the individual is human.

45. The computer storage media of any one of claims 42-44, wherein the hematological cell parameter (e.g., MDW), the anion gap, the measure of PCT, the measure of CRP, the measure of ALK, the measure of AST, the measure of calcium, or the O2 saturation, or a combination thereof are determined from the sample.

46. The computer storage media of any one of claims 42-45, further comprising: communicatively coupling with a database; and retrieving, from the database values associated with the hematological cell parameters(e g., MDW), the anion gap, the measure of PCT, the measure of CRP, the measure of ALK, the measure of AST, the measure of calcium, or the O2 saturation.

47. The computer storage media of any one of claims 42-46, wherein the trained model is trained using a log loss function to optimize true positives for bacterial or viral infection.

48. The computer storage media of any one of claims 42-47, wherein the input data set for the individual additionally includes an eosinophil count, neutrophil count, and erythrocyte sedimentation rate (ESR) or a combination thereof.

49. The computer storage media of any one of claims 42-48, wherein the input data set for the individual additionally includes at least one of red blood cell (RBC) count, basophil count, lymphocyte count, a measure of hemoglobin, white blood cell (WBC) count, red cell distribution width (RDW), monocyte count, a measure of mean platelet volume (MPV), platelet count, or neutrophil to lymphocyte ratio.

50. The computer storage media of any one of claims 42-49, wherein volume conductivity scatter parameters of monocytes are measured for a portion of the sample to generate diagnostic information.

51. The computer storage media of any one of claims 42-49, wherein light scatter parameters of monocytes are measured for a portion of the sample to generate diagnostic information.

52. A system comprising: a light scatter detection sensor; one or more processors; non-transitory computer storage media communicating coupled to the one or more processors, the computer storage media storing instructions that when executed by the one or more processors cause the one or more processors to perform a process comprising:generating a hematological cell parameter for an individual based on measurements for a sample; obtaining, clinical measurements for the sample, wherein the clinical measurements comprise: an anion gap, a measure of procalcitonin (PCT), a measure of C-reactive protein (CRP), alkaline phosphatase (ALK), aspartate aminotransferase (AST), calcium, or oxygen (O2) saturation, or a combination thereof; generating an input data set for the individual including the MDW value and the clinical measurements; providing the input data set as input to a trained model configured to generate an output value; comparing the output value of the trained model to predetermined threshold values to determine probabilities associated with a bacterial infection, a viral infection, or a combination or absence thereof.

Images