AUTHOR=Chen Ting , Zhang Xuefeng , Yu Qunfeng , Yang Qin , Yuan Lingmin , Tong Fei 

TITLE=Construction and clinical visualization application of a predictive model for mortality risk in sepsis patients based on an improved machine learning model

JOURNAL=Frontiers in Physiology

VOLUME=Volume 16 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/physiology/articles/10.3389/fphys.2025.1560659

DOI=10.3389/fphys.2025.1560659

ISSN=1664-042X

ABSTRACT=ObjectiveTo explore the construction and clinical visualization application of a mortality risk prediction model for sepsis patients based on an improved machine learning model.MethodsThis retrospective study analyzed 1,050 sepsis patients admitted to Longyou County People’s Hospital between January 2010 and August 2023. Patients were divided into a survival group (n = 877) and a death group (n = 173) based on their 30-day mortality status. Clinical and laboratory data were collected and used as feature variables. A Self-Weighted Self-Evolutionary Learning Model (SWSELM) was developed to identify independent risk factors for sepsis mortality and to create a visualization system for clinical application.ResultsThe improved algorithm significantly outperformed other algorithms on 23 standard test functions. The SWSELM model achieved ROC-AUC and PR-AUC values of 0.9760 and 0.9624, respectively, on the training set, and 0.9387 and 0.9390, respectively, on the test set, both significantly higher than those of three other prediction models. The SWSELM model identified 10 important features, with multivariate logistic regression retaining five variables: B-type Natriuretic Peptide Precursor (NT-proBNP), Lactate, Albumin, Oxygenation Index, and Mean Arterial Pressure (MAP) (OR = 4.889, 3.770, 3.083, 1.872, 1.297), consistent with the top five features selected by the SWSELM model.ConclusionNT-proBNP, Lactate, Albumin, Oxygenation Index, and Mean Arterial Pressure are independent risk factors for mortality in sepsis patients. This study successfully created a self-evolutionary prediction model using machine learning methods, demonstrating significant clinical application potential and value for broader implementation.