Predicting Patient-Specific Single-Cell Parameters in Computational Cardiac Models Using Machine Learning.

Sudden cardiac death (SCD) causes >4 million deaths worldwide annually. A computational cardiac simulation that is personalized with patient-specific physiological parameters might provide more accurate representations of potential SCD-causing electrical activity and suggest therapeutic insights. However, experimentally obtaining cellular patient-specific parameters is difficult and time-consuming. To circumvent this limitation, we describe a machine learning approach to learn cellular parameters from action potential (AP) and intracellular Ca+(Cai) data.