JHU Selects ADVANCE ECG Deep Learning Project for Discovery Award
JHU Selects ADVANCE ECG Deep Learning Project for Discovery Award
Johns Hopkins University’s Discovery Awards encourage faculty from various disciplines to collaborate in addressing multifaceted challenges and pushing the boundaries of understanding. In 2024, 44 Discovery Awards were given to interdisciplinary faculty teams, chosen from 286 proposals. The ADVANCE proposal, entitled “Deep Learning Applications in ECG Analysis for Atrial Fibrillation Patients: A New Horizon in Diagnosis and Monitoring” led by PI Eugene Kholmovski and in collaboration with members of the Trayanova lab, was one of the 44 proposals selected for the award.
The project can be briefly summarized as follows:
Atrial fibrillation (AF), the most common cardiac arrhythmia, significantly compromises patient quality of life.This condition leads to an increase in hospitalizations, stroke and mortality rates, exerting a considerable financial burden on healthcare system due to the extensive costs associated with its management. Current diagnostic techniques for these co-morbidities, such as electro-anatomical mapping and late-gadolinium-enhanced magnetic resonance imaging (LGE-MRI) for evaluating the atrial substrate, echocardiography for evaluating valve diseases and polysomnography for OSA screening, are effective but encounter challenges in widespread application. These challenges arise partly from the invasive nature of some of these methods and partly from limitations in accessibility and practicality, attributed to their complexity and significant resource demand. There is a growing need for non-invasive, efficient, and widely accessible diagnostic tools, especially considering the high prevalence and public health burden of AF and its associated co-morbidities.
In this context, the integration of artificial intelligence (AI) and deep learning with electrocardiograms (ECGs), a non-invasive and widely accessible diagnostic tool, presents a promising avenue to address these challenges. The goal of this project is to leverage the latest advancements in AI and deep learning for the analysis of ECGs and clinical covariates to create a diagnostic tool that is not only precise and efficient but also cost-effective and easily accessible, specifically for the screening and monitoring of AF patients.
The project introduces pioneering approaches through the application of deep learning to ECG analysis, aimed at enhancing patient care in AF by:
- Predicting the atrial substrate and monitoring the progression of atrial cardiomyopathy in AF patients, thereby offering a novel diagnostic method for early detection and longitudinal tracking of disease evolution.
- Prognosticating the success of pulmonary vein isolation in AF patients. This will enable the stratification of patients who could benefit from additional substrate modification, thus promoting the development of personalized ablation strategies and potentially improving clinical outcomes.
- Identifying and stratifying co-morbidities in AF patients, such as obstructive sleep apnea, diabetes mellitus, and valve diseases, which could lead to more comprehensive patient assessments and integrated management strategies.
Dr. Mohsen, a visiting scholar from Hospital Porz am Rhein, Germany, has provided essential data and insights from his institution, reflecting our project’s commitment to leveraging multi-center data and expertise for advancements in cardiovascular health. Dr. Kholmovski and his team within ADVANCE are excited to begin this work with the support of the Discovery Award, for which they are grateful to JHU, and the assistance of Dr. Mohsen.
