This study developed a novel deep learning (DL) approach combining neural networks and survival analysis to predict patient-specific survival curves for patients with ischemic heart disease, using contrast-enhanced cardiac magnetic resonance images and clinical covariates. The DL-predicted survival curves accurately predicted survival up to 10 years, including uncertainty estimates. The model achieved concordance indexes of 0.83 and 0.74 and 10-year integrated Brier scores of 0.12 and 0.14 in internal validation and independent test sets, respectively, outperforming standard survival models using only clinical covariates. This technology could improve clinical decision-making by providing accurate, generalizable predictions of arrhythmic death probabilities.

Dan M. Popescu, Julie K. Shade, Changxin Lai, Konstantinos N. Aronis, David Ouyang, M. Vinayaga Moorthy, Nancy R. Cook, Daniel C. Lee, Alan Kadish, Christine M. Albert, Katherine C. Wu, Mauro Maggioni, Natalia A. Trayanova