Mathematical models and artificial intelligence for studying the ventricular dyssynchrony and improving the effectiveness of cardiac resynchronization therapy

The review is devoted to the current status of artificial intelligence (AI), mathematical modeling (MM) and their combination in the analysis of cardiac dyssynchrony mechanisms and its treatment using cardiac resynchronization therapy (CRT). Recent articles and reviews demonstrate the high promise of AI and MM in personalized medicine, but also identify existing obstacles to their implementation in clinical practice. The works discussed are devoted to a number of topical problems of clinical cardiology as follows: analysis of phenotypes of patients with cardiac dyssynchrony, search for novel prognostic factors of CRT effectiveness, pacing optimization, creation of highly accurate predictive models of response to CRT. For the first time, a review of studies is given that use combined approaches of mechanistic MM and AI. Such approaches break new ground for the application of personalized heart models both for generating realistic synthetic data (digital twins) on which AI models are trained, and as predictors that, along with clinical signs, are used in trained prognostic AI models to improve the accuracy of personalized diagnostics, predict the effectiveness and optimize treatment. The review consists of three sections focused on studies using AI, personalized MM of heart ventricles, and combined approaches (MM+AI).