Modeling ischemia with finite elements and automated machine learning

Abstract

© 2018 Elsevier B.V. The main purpose of this study was to noninvasively detect and localize ischemic cardiac disease using the finite element method (FEM) in combination with machine learning approach. The forward FEM simulations of cardiac ischemia in different heart segments enabled the creation of a virtual database which consisted of corresponding body surface potentials. Two sets of experiments were performed on the database in order to select the best method for determining the existence of ischemia and then to predict the location of ischemia. Using Auto-WEKA and R caret package, several machine learning algorithms were tested. The best first phase model returned the classification accuracy of 95.3%, while the best second phase model determined the correct ischemia location with 95% accuracy. Considerable modeling and computational time are needed to create a training database and perform training, but once trained, the models will instantly return results. Thus, the main advantage of the proposed approach to ischemic detection and localization is a real-time availability of results and a novel, two-phase design which guides the selection of an adequate treatment.