Related Experiment Videos
Development and external validation of the HCH and HPMS prognostic indices for sepsis: a retrospective model
Chengcheng Gao1,2, Haiyue Zhang1, Rui Zhang3
1Department of Health Statistics, Ministry of Education Key Lab of Hazard Assessment and Control in Special Operational Environment, School of Public Health, Fourth Military Medical University, Xi'an, Shaanxi, 710032, China.
BMC Medical Informatics and Decision Making
|June 3, 2026
Summary
A new algorithm identified novel sepsis indicators (HCH and HPMS) that improve mortality prediction over traditional scores. These indices offer better clinical stratification and interpretability for sepsis management.
Area of Science:
- * Critical Care Medicine
- * Computational Biology
- * Biomedical Engineering
Background:
- * Sepsis presents pathological heterogeneity, challenging traditional scoring systems' sensitivity, dynamic characterization, and interpretability.
- * Developing novel composite indices from routine indicators using advanced algorithms can enhance sepsis assessment and prognosis.
- * This study aimed to create and validate new composite indices for predicting sepsis mortality.
Purpose of the Study:
- * To develop and externally validate novel composite indices for sepsis mortality prediction.
- * To address the limitations of traditional scoring systems in balancing sensitivity, dynamic progression, and interpretability.
- * To leverage rheological properties and multi-objective optimization for improved sepsis assessment.
Main Methods:
- * Developed the Multi-Objective Non-Newtonian Fluid Optimization (MONNF) algorithm using data from eICU and MIMIC-IV databases.
- * Integrated dynamic viscoelastic regulation, multi-target constraints, and medical knowledge to screen indicators capturing sepsis pathophysiology.
- * Established an Under-sampling Synchronous Evolutionary Ensemble (USEE) prediction model for cross-center validation.
Main Results:
- * Identified the Hypoxia-Coagulation-Hemoglobin (HCH) index (lactate, INR, hemoglobin) and Hepatic-Pulmonary-Metabolic Synergistic (HPMS) index (total bilirubin, PaO₂/FiO₂, bicarbonate, albumin).
- * HCH showed superior 28-day mortality prediction (ROC-AUC=0.67) compared to SOFA (ROC-AUC=0.63).
- * The USEE model achieved optimal predictive performance in validation (ROC-AUC=0.84, PR-AUC=0.71), indicating strong generalizability.
Conclusions:
- * A novel algorithm based on non-Newtonian fluid properties successfully identified composite sepsis indicators.
- * These new indices characterize core pathological features, offering a simplified clinical stratification.
- * The study provides an innovative solution for sepsis prognostic assessment, improving efficiency, sensitivity, and interpretability.
