The Study on the Prognostic Assessment Value of Pan-Immune-Inflammation Value (PIV) in Patients with Sepsis

Background Sepsis remains a leading cause of mortality in intensive care units worldwide. The pan-immune-inflammation value (PIV), a composite biomarker derived, has shown prognostic value in oncology, but its role in sepsis has not been well established. Objective To evaluate the prognostic significance of PIV for 28- and 90-day mortality in adult sepsis patients and to develop and externally validate PIV-based predictive models. The study also aimed to explore dynamic PIV trajectories and their associations with early organ dysfunction, including acute kidney injury (AKI) and acute respiratory distress syndrome (ARDS). Methods Adult sepsis patients from the MIMIC-IV and eICU databases were analyzed. PIV was assessed using survival analysis and multiple machine learning algorithms (Random Baseline, Logistic Regression, Gradient Boosting Classifier, AdaBoost, XGBoost, LightGBM) after LASSO regression–based feature selection and hyperparameter optimization. Model performance was evaluated with ROC curves and SHAP analyses. Group-based trajectory modeling (GBTM) was used to identify PIV dynamic patterns over the first 7 ICU days and assess their associations with AKI and ARDS. Results Elevated PIV values were significantly associated with higher 28-day and 90-day all-cause mortality (P < 0.05). LASSO regression confirmed PIV as a key prognostic feature. PIV-based models—especially Gradient Boosting Classifier and XGBoost—showed strong discrimination for short- and intermediate-term mortality. Four PIV trajectories (Traj-1 to Traj-4) were identified. Compared with Traj-1, Traj-3 exhibited a markedly increased risk of ARDS (AUC = 0.611) and AKI, followed by Traj-2, while Traj-4 showed no significant difference. Conclusions PIV is an effective immune-inflammatory biomarker for predicting sepsis outcomes. PIV-based machine learning models demonstrate promising accuracy for individualized mortality prediction. Moreover, dynamic PIV trajectories correlate with early organ dysfunction, supporting their use in early risk stratification and precision management in sepsis.