Prognostic Value of SIRI in Sepsis: A Retrospective Study and Machine Learning-Based Model Development

Area of Science:

• Critical Care Medicine
• Sepsis Pathophysiology
• Prognostic Biomarkers

Background:

• Sepsis poses a significant global health challenge with high mortality rates.
• Accurate prognostic tools are crucial for timely clinical decision-making in sepsis management.
• The Systemic Inflammation Response Index (SIRI) has emerged as a promising biomarker for evaluating sepsis prognosis.

Purpose of the Study:

• To investigate the predictive value of SIRI for 28-day outcomes in sepsis patients.
• To develop and validate a prognostic model for predicting 28-day mortality in sepsis.
• To compare the predictive performance of SIRI against established scoring systems like APACHE II and SOFA.

Main Methods:

• Retrospective analysis of adult sepsis patients from two medical centers.
• Utilized restricted cubic splines and ROC curve analysis to assess SIRI's predictive capability.
• Employed Cox regression, Kaplan-Meier curves, Boruta algorithm, LASSO, and logistic regression to build and validate a prognostic model.
• External validation was performed using data from a separate hospital.

Main Results:

• SIRI demonstrated a nonlinear increasing trend with mortality risk and superior predictive capability compared to APACHE II and SOFA scores.
• Higher SIRI levels were significantly associated with poorer 28-day prognosis (p<0.001).
• The developed prognostic model, incorporating SIRI and other factors (BUN, age, P, Lac, MV), showed high predictive performance (AUCs: 0.851-0.908) across training and validation sets.

Conclusions:

• SIRI is a valuable and independent predictor of 28-day prognosis in sepsis patients.
• The developed prognostic model incorporating SIRI offers high accuracy and clinical utility for predicting sepsis mortality.
• This model can aid clinicians in risk stratification and management of sepsis patients.