Risk model for predicting mortality risk in patients with cerebral infarction: a cohort study using the MIMIC-IV database

Cerebral infarction represents a significant health threat, causing mortality and long-term disability. This study investigated prognostic variables associated with cerebral infarction and developed a risk model to predict mortality, thereby offering new strategies to enhance patient care and improve overall outcomes. Participants were selected from the Medical Information Mart for Intensive Care IV database, covering the period 2008–2019.

The study population comprised patients with cerebral infarction admitted to the intensive care unit. Univariate Cox regression analysis, proportional hazards assumption testing, machine learning (least absolute shrinkage and selection operator regression), and multivariate Cox regression analysis were performed to identify prognostic variables. Based on these variables, a risk model was developed and subsequently evaluated.

After excluding ineligible individuals, 109 patients with cerebral infarction were included. The anion gap [hazard ratio (HR) = 1.1040, 95% confidence interval (CI) = 1.0070–1.2110, P = 0.0356], platelet count (HR = 0.9965, 95% CI = 0.9939–0.9990, P = 0.0072), and glucose concentration (HR = 1.0050, 95% CI = 1.0000–1.0090, P = 0.0483) were identified as prognostic variables. A risk model utilizing these prognostic variables classified patients into high- and low-risk cohorts, with high-risk patients displaying lower survival probabilities.

Receiver operating characteristic curve analysis demonstrated the strong predictive performance of this risk model, with all areas under the curve exceeding 0.7. The study identified the anion gap, platelet count, and glucose concentration as prognostic variables for cerebral infarction. The findings underscore the potential of the risk model to enhance clinical decision-making and management for patients with cerebral infarction.

We would like to express our sincere gratitude to all individuals and organizations who supported and assisted us throughout this research. Special thanks to the following authors: Zhang Jinzhi. In conclusion, we extend our thanks to everyone who has supported and assisted us along the way.

Without your support, this research would not have been possible. The research reported in this project was generously supported by Supported by Chengdu Science and Technology Program under grant agreement number 2024-YF05-01785-SN. Department of Neurology, School of Medicine, Sichuan Provincial People’s Hospital, University of Electronic Science and Technology of China, Chengdu, 610072, Sichuan Province, China Department of Neurology, The Six People’s Hospital of Chengdu, Chengdu, 610051, Sichuan Province, China Department of Neurology, The People’s Hospital of Jian Yang City, Chengdu, 641400, Sichuan Province, China The authors declare no competing interests.