Abstract:Abstract Sepsis is characterised by a dysfunctional host response to infection culminating in life-threatening organ failure that requires complex patient management and rapid intervention. Timely diagnosis of the underlying cause of sepsis is crucial, and identifying those at risk of complications and death is imperative for triaging treatment and resource allocation. Here, we explored the potential of explainable machine learning models to predict mortality and causative pathogen in sepsis patients. By using a modelling pipeline employing multiple feature selection algorithms, we demonstrate the feasibility to identify integrative patterns from clinical parameters, plasma biomarkers and extensive phenotyping of blood immune cells. Whilst no single variable had sufficient predictive power, models that combined five and more features showed a macro area under the curve (AUC) of 0.85 to predict 90 day mortality after sepsis diagnosis, and a macro AUC of 0.86 to discriminate between Gram-positive and Gram-negative bacterial infections. Parameters associated with the cellular immune response contributed the most to models predictive of 90 day mortality, most notably, the proportion of T cells among PBMCs, together with expression of CXCR3 by CD4+ T cells and CD25 by mucosal-associated invariant T (MAIT) cells. Frequencies of Vδ2+ γδ T cells had the most profound impact on the prediction of Gram-negative infections, alongside other T cell-related variables and total neutrophil count. Overall, our findings highlight the added value of measuring the proportion and activation patterns of conventional and unconventional T cells in the blood of sepsis patients in combination with other immunological, biochemical and clinical parameters.

What problem does this paper attempt to address?

The problem this paper attempts to address is: by integrating clinical parameters, plasma biomarkers, and extensive phenotypic analysis of blood immune cells, using interpretable machine learning models to predict the 90-day mortality and pathogenic bacteria types in sepsis patients. Specifically, the study aims to: 1. **Predict 90-day mortality**: Develop models that can accurately predict the 90-day mortality risk in sepsis patients. 2. **Differentiate pathogenic bacteria types**: Identify combinations of biomarkers that can distinguish between Gram-positive and Gram-negative bacterial infections. ### Background and Importance Sepsis is a life-threatening syndrome caused by infection, characterized by organ dysfunction, requiring complex patient management and timely intervention. Timely diagnosis of the underlying cause of sepsis is crucial, especially to identify patients at risk of complications and death, to allocate treatment resources appropriately. Currently used severity scoring tools (such as APACHE and SOFA), although useful for auditing and guiding care, perform poorly in predicting in-hospital mortality and fail to capture the complex immune and metabolic mechanisms leading to sepsis pathology. ### Research Methods Researchers used various feature selection algorithms, combining routine clinical data, plasma biomarkers, and immune cell phenotypes, to construct multiple machine learning models. These models aim to identify comprehensive patterns that can predict 90-day mortality and differentiate between Gram-positive and Gram-negative bacterial infections. ### Key Findings - **Limited predictive ability of single variables**: No single variable had sufficient predictive ability. - **Effectiveness of multivariable models**: Models combining 5 or more features showed high predictive performance, with a macro AUC of 0.85 for 90-day mortality and a macro AUC of 0.86 for differentiating Gram-positive and Gram-negative bacterial infections. - **Importance of immune cell features**: Parameters related to cellular immune response contributed most to predicting 90-day mortality, particularly the proportion of T cells, CXCR3 expression on CD4+ T cells, and CD25 expression on mucosal-associated invariant T (MAIT) cells. - **Role of specific T cell subsets**: The frequency of Vδ2+ γδ T cells was particularly significant in predicting Gram-negative bacterial infections, with other T cell-related variables and total neutrophil count also playing important roles. ### Conclusion The study results emphasize the importance of measuring the proportions and activation patterns of conventional and unconventional T cells in the blood of sepsis patients, combined with other immunological, biochemical, and clinical parameters. These comprehensive models help prioritize and guide patient care in resource-limited settings.

Conventional and unconventional T cell responses contribute to the prediction of clinical outcome and causative bacterial pathogen in sepsis patients

Machine Learning Model for Early Prediction of Sepsis Outcomes Based on Immune Response

Deep Learning-Based Clustering Robustly Identified Two Classes of Sepsis with Both Prognostic and Predictive Values

Machine Learning Algorithms Identify Pathogen-Specific Biomarkers of Clinical and Metabolomic Characteristics in Septic Patients with Bacterial Infections

An Integrated Clinico-Metabolomic Model Improves Prediction of Death in Sepsis

A Signature of Immune-Related Genes Correlating with Clinical Prognosis and Immune Microenvironment in Sepsis

Classification of Patients With Sepsis According to Immune Cell Characteristics: A Bioinformatic Analysis of Two Cohort Studies

Machine Learning Identifies Complicated Sepsis Course and Subsequent Mortality Based on 20 Genes in Peripheral Blood Immune Cells at 24 H Post-ICU Admission

Development and validation of a machine learning-based interpretable model for predicting sepsis by complete blood cell parameters

The application of metagenomics, radiomics and machine learning for diagnosis of sepsis

An integrative model with HLA‐DR, CD64, and PD‐1 for the diagnostic and prognostic evaluation of sepsis

A Machine learning model for predicting sepsis based on an optimized assay for microbial cell-free DNA sequencing

Predicting sepsis severity at first clinical presentation: The role of endotypes and mechanistic signatures

A Comprehensive Analysis of the Immune Microenvironment in Peripheral Blood Promoted Precise Evaluation of the Prognosis of Sepsis and Sepsis-Induced Acute Respiratory Distress Syndrome Patients: A Machine Learning , Multicohort Study

Peripheral PD-1+NK cells could predict the 28-day mortality in sepsis patients

Exploring Immune-Related Gene Expression Up to the First 24-Hour for Predicting Sepsis Outcomes Based on Comprehensive Bioinformatics Analysis and Machine Learning

Identifying Predictors Associated with Risk of Death or Admission to Intensive Care Unit in Internal Medicine Patients with Sepsis: A Comparison of Statistical Models and Machine Learning Algorithms

Dynamic changes in human single-cell transcriptional signatures during fatal sepsis

Interpretable Machine Learning for Early Prediction of Prognosis in Sepsis: A Discovery and Validation Study

An immune genes signature for predicting mortality in sepsis patients

Evaluation of a combination “lymphocyte apoptosis model” to predict survival of sepsis patients in an intensive care unit