Machine learning-based decision support system for the prognostication of neurological outcome in the successfully resuscitated OHCA patient

Abstract Background This study uses machine learning and multi-center registry data for analyzing the determinants of favorable neurological outcome in the out-of-hospital cardiac arrest (OHCA) patient and developing its decision support systems for various subgroups. Methods Data came from Korean Cardiac Arrest Research Consortium registry with 2679 OHCA patients aged 18 or more with the return of spontaneous circulation (ROSC). The dependent variable was favorable neurological outcome (Cerebral Performance Category scores 1–2) and 68 independent variables were included, e.g., first monitored rhythm, in-hospital cardiopulmonary resuscitation (CPR) duration and post-ROSC pH. The random forest was used for identifying major determinants of favorable neurological outcome and developing its decision support systems for various subgroups stratified by major variables. Results Based on random forest variable importance, major determinants of OHCA outcome were in-hospital CPR duration (0.0824), in-hospital electrocardiogram on emergency room arrival (0.0692), post-ROSC pH (0.0579), prehospital ROSC before emergency room arrival (0.0565), coronary angiography (0.0527), age (0.0415), first monitored rhythm (EMS) (0.0402), first monitored rhythm (community) (0.0401), early coronary angiography within 24 hours (0.0304) and scene arrival to CPR stop (0.0301). It was also found that patients can be divided to 6 subgroups in terms of prehospital ROSC and first monitored rhythm (EMS) and that a decision tree can be developed as a decision support system for each subgroup to find its effective cut-off points regarding in-hospital CPR duration, post-ROSC pH, age and hemoglobin. Conclusions We identified the major determinants of favorable neurological outcome in successfully resuscitated OHCA patients using machine learning. This study demonstrated the strengths of the random forest as an effective decision support system for each stratified subgroup (prehospital ROSC and first monitored rhythm by EMS) to find its own optimal cut-off points for major in-hospital variables (in-hospital CPR duration, post-ROSC pH, age and hemoglobin).