National Heart, Lung, and Blood Institute 1R21HL175632-01A1

Evaluating the efficacy of computer aided exercise stress ECG reader (CAESER) for automated diagnosis of coronary artery disease

This NIH-funded project aims to develop and evaluate CAESER, a machine learning tool for automated diagnosis of coronary artery disease using exercise stress ECG data, leveraging large datasets and deep learning to improve diagnostic accuracy and provide low-cost, continuous risk monitoring.

Description

Primary goal of this project is to significantly elevate the positive predictive value (PPV) and sensitivity of exercise stress electrocardiography (ESE) in assessing obstructive coronary artery disease (CAD), defined as > 50% stenosis identified by invasive coronary angiogram (ICA) across pretest cardiac risk spectrum, by leveraging large, harmonized ESE dataset correlated with ICA. CAESER (Computer Aided Exercise Stress ECG Reader) can automatically analyze ESE and provide statistical likelihood of CAD by adopting an expert-AI collaborative precision cardiology approach, where gender-specific baseline STD are captured by a digital twin, GeMREM, while the continuous comparison with baseline is achieved through transformer-based deep learning architecture. The machine learning evaluation framework developed in this project can lead to a lightweight automatic CAD diagnosis approach to be integrated with mobile ECG monitoring and provide low-cost CAD risk monitoring that has uniform performance across gender, racial demographics, socio-economic status, and co-morbidities.

View original record