This study aims to propose a clinically interpretable ensemble approach for the timely and accurate prediction of cardiac arrest (CA) within 24 hours, regardless of patient heterogeneity, including variations across different populations and intensive care unit (ICU) subtypes.
Additionally, we conducted patient-independent evaluations to emphasize the model's generalization performance and analyzed interpretable results that can be readily adopted by clinicians in real-time. Patients' data from two database (MIMIC-IV and eICU-CRD) were used.
The proposed method outperformed conventional approaches across different cohort populations in both MIMIC-IV and eICU-CRD. Additionally, it achieved higher accuracy than baseline models for various ICU subtypes within both databases.
This study's novel framework for learning unique features provides stable predictive power across different ICU environments. Most of the interpretable global information reveals statistical differences between CA and non-CA groups, demonstrating its utility as an indicator for clinical decisions.
【MORE】