A Least Absolute Shrinkage and Selection Operator (LASSO)-derived AAGC Model for In-hospital Mortality Prediction in Sepsis Incorporating Age, APACHE II Score, Glasgow Coma Scale, and Creatinine: A Prospective Study.

Background and aims: Sepsis remains a major global cause of critical illness and death. Existing scores like Acute Physiology and Chronic Health Evaluation II (APACHE II), Sequential Organ Failure Assessment (SOFA), and quick Sequential Organ Failure Assessment (qSOFA) vary widely, highlighting the need for earlier, reliable risk identification. We aimed to identify key predictors of mortality among inpatients with sepsis and develop a simplified data-driven predictive model using the least absolute shrinkage and selection operator (LASSO) method. Patients and methods: A prospective cohort study was carried out involving 150 adult patients diagnosed with sepsis and admitted to a tertiary care hospital in India between November 2024 and April 2025. Demographic, clinical, and biochemical data were recorded at admission. Least absolute shrinkage and selection operator regression was used for variable selection, followed by multivariable logistic regression to develop the final model. Model performance was assessed through receiver operating characteristic (ROC) analysis, calibration plots, and decision curve analysis. Results: Of the 150 patients, 86 (57.3%) died during hospitalization. Least absolute shrinkage and selection operator regression identified four key predictors: Age, APACHE II score, Glasgow Coma Scale (GCS) and creatinine, forming the AAGC model. The model demonstrated excellent discrimination (AUC: 0.95, 95% CI: 0.92-0.99) and strong calibration (mean absolute error: 0.011; 90th percentile: 0.026). Decision curve analysis showed greater net benefit across threshold probabilities of 0.1-0.8 compared with "treat-all" or "treat-none" approaches. High respiratory rate (RR), culture-negative sepsis, respiratory failure, and septic shock were also independently associated with mortality. Conclusion: This LASSO-based AAGC model offers a robust, interpretable tool for early mortality prediction in sepsis, thereby supporting timely interventions and improving patient outcomes. Further external validation of this model may confirm its generalizability and its clinical utility. How to cite this article: Gulia H, Suhag M, Bagal N, Deshwal PR, Kumar S, Joshi RK. A Least Absolute Shrinkage and Selection Operator (LASSO)-derived AAGC Model for In-hospital Mortality Prediction in Sepsis Incorporating Age, APACHE II Score, Glasgow Coma Scale, and Creatinine: A Prospective Study. Indian J Crit Care Med 2026;30(5):379-388.