Prognostic value of the pan-immune-inflammation value for mortality in sepsis-induced coagulopathy: a Medical Information Mart for Intensive Care study.

Background: Sepsis is a life-threatening condition characterized by organ dysfunction caused by a dysregulated host response to infection. Its associated coagulopathy, known as sepsis-induced coagulopathy (SIC), significantly increases mortality risk. The pan-immune-inflammation value (PIV), a composite biomarker reflecting systemic immune and inflammatory status, has been linked to prognosis in various diseases. Objectives: This study aimed to evaluate the prognostic significance of PIV in patients with SIC and to develop predictive models accordingly. Methods: This retrospective study utilized data from the Medical Information Mart for Intensive Care IV database and included 4554 patients diagnosed with sepsis. Patients were stratified into high- and low-PIV groups based on the median PIV, and clinical characteristics were compared between groups. Kaplan-Meier survival analysis and Cox regression were employed to assess the association between PIV and patient outcomes. Least absolute shrinkage and selection operator regression was used to identify key variables for constructing a nomogram model. Additionally, machine learning algorithms, including random forest, were applied to build and validate predictive models. Results: Patients in the high-PIV group had significantly higher 30-day and 90-day mortality rates. Kaplan-Meier analysis showed that patients with lower PIVs had markedly better survival, and a nonlinear positive correlation was observed between PIV and mortality risk. Least absolute shrinkage and selection operator regression identified 8 key variables, including Acute Physiology Score III, lactate, red cell distribution width, mean corpuscular volume, acute kidney injury, and continuous renal replacement therapy. The nomogram based on these variables achieved areas under the receiver operating characteristic curve of 0.84 and 0.87 in the training and validation cohorts, respectively. Among machine learning models, the random forest algorithm exhibited the best predictive performance, with areas under the curve of 0.837 and 0.947 in the training and validation sets, respectively. External validation using a real-world cohort from Xingtai People's Hospital further confirmed the association between elevated PIV and increased mortality and SIC, with consistent survival trends and nonlinear patterns observed in both Kaplan-Meier and restricted cubic spline analyses. Conclusion: To our knowledge, this study is the first to incorporate PIV into the prognostic assessment of patients with SIC. The development of a visual nomogram and machine learning-based models provides clinicians with practical tools for early identification of patients at high risk for SIC, potentially aiding in the optimization of treatment strategies.