Machine Learning Prediction of Mortality Following Cardiac Surgery: Analysis of a Large National Database

This study compared in-hospital mortality risk prediction after cardiac surgery using the EuroSCORE II system with results generated by several machine-learning techniques. The dataset contained 227,087 adult patients, among whom there were 6,258 deaths (2.76 percent). Machine learning techniques showed some statistical improvements over EuroSCORE II, but the clinical impact of this was modest. The addition of other risk factors in future studies may result in further improvements.