Association between the nutritional inflammation index and mortality among patients with sepsis: insights from traditional methods and machine learning-based mortality prediction.

backgroundSepsis is a life-threatening condition characterized by dysregulated immune responses and metabolic disturbances. The albumin-to-neutrophil-lymphocyte ratio (ANLR) is a novel composite biomarker integrating nutritional and inflammatory status. However, its prognostic significance in sepsis remains unclear. This study aims to evaluate the association between ANLR and mortality in sepsis patients using both traditional statistical methods and machine learning models.

methodsA retrospective cohort study was conducted using the MIMIC-IV (v3.1) database. In this study, 6,288 patients diagnosed with sepsis and admitted to the ICU were analyzed, with participants stratified into quartiles according to their ANLR measurements. The primary endpoint was set as 30-day mortality, while 90-day mortality served as a secondary outcome. The association between ANLR and mortality was investigated through Kaplan-Meier survival analysis, Cox regression, and restricted cubic spline (RCS) modeling. Furthermore, the contribution of ANLR relative to other predictors was evaluated by developing machine learning models, with SHapley Additive exPlanations (SHAP) employed to determine variable importance.

resultsA higher ANLR was independently associated with improved survival. In the fully adjusted Cox model, elevated ANLR predicted a lower risk of mortality at 30 days (HR 0.68, 95% CI 0.59-0.79, p < 0.001) and at 90 days (HR 0.85, 95% CI 0.76-0.94, p = 0.002). Machine learning analysis identified ANLR as the second most important variable influencing sepsis mortality. ANLR demonstrated superior predictive ability (AUC 0.66) compared to traditional markers, including SOFA, NLR, and albumin.

conclusionsANLR is a robust and independent predictor of sepsis-related mortality, outperforming conventional biomarkers. Incorporating ANLR into routine clinical workflows could improve risk assessment and facilitate individualized treatment strategies for patients with severe sepsis. Further prospective studies are needed to validate these findings and explore potential therapeutic implications.