Precision Prediction of Right Ventricular Size, Function, and Outcomes in Patients with Repaired Tetralogy of Fallot - PROJECT SUMMARY: Dr. Son Duong (PI) is a clinician-scientist specializing in pediatric cardiology and advanced cardiovascular imaging at the Icahn School of Medicine at Mount Sinai (ISMMS). The primary objective of this application is to support the PI’s career development into an independent investigator of the translational application of artificial intelligence to cardiovascular imaging and patient risk prediction in patients with tetralogy of Fallot and other congenital heart disease. Patients with repaired tetralogy of Fallot (rTOF) are at risk of are at increased risk of heart failure, arrhythmia, and sudden cardiac death, but current methods of clinical risk stratification for poor outcome are inaccurate and limited in the feature set they incorporate. Accurate assessment and monitoring of right ventricular (RV) volumes and ejection fraction (EF) are critical for effective management of patients with rTOF, and thus cardiac MRI (cMRI) is recommended every 1-3 years in adulthood because traditional methods for quantitative RV assessment are limited. Frequent cMRI is a patient and system burden which may limit care. This proposal will leverage deep learning tools for the following specific aims: Develop and validate a prediction tool of RV volumes and EF from 12-lead electrocardiogram waveforms and two-dimensional echocardiographic video (Aim 1); and incorporate electrocardiogram, echocardiogram, and electronic health record data into a multimodal clinical risk stratification tool (Aim 2). The PI’s training plan focuses on coursework, tutorials, and seminars/workshops in four key areas: (1) statistical machine learning, (2) deep learning, (3) patient-oriented clinical risk prediction, and (4) career development. To meet these research and career development goals, the PI has assembled an expert cross-disciplinary mentorship team consisting of primary mentor Dr. Girish Nadkarni MD, MS (Professor of Medicine and Division Chief of Data Driven and Digital Medicine), a clinical informaticist with expertise in machine- and deep learning; co-mentors Dr. Bruce Gelb MD (Professor of Pediatrics and Genetics and Genomic Sciences and Dean of Child Health Research), a pediatric cardiologist with extensive expertise in study design and scientific career mentorship; Dr. Hayit Greenspan PhD (Professor of Diagnostic, Molecular and Interventional Radiology, and Director of Artificial Intelligence in Imaging), a computer scientist with expertise in deep learning in medical image analysis; and Dr. Brett Anderson (Associate Professor of Pediatrics), an expert of multicenter outcomes research in pediatric cardiology. The ISMMS Department of Pediatrics has a strong track-record of NIH funding and successful mentorship to scientific independence. They will provide the PI with access to cutting-edge computational power, clinical data infrastructure, and robust collaborative and educational platforms, as well as protected time to ensure the PI’s success. The results from these research aims will set the foundation for future R01 studies for multicenter validation, expansion into other congenital heart disease populations, and translation of these findings to clinical practice in a prospective fashion.
