Academia Sinica Logo

Electrocardiography swiftly identifies cardiac disorders, spanning structural anomalies, arrhythmias, and myocardial ischemia. Our team pioneered an integrated AI-enabled ECG platform, leveraging extensive annotated data and cutting-edge deep learning techniques. This innovative system accurately identifies over 50 conditions, encompassing both acute cardiac crises and chronic ailments, through a single 12-lead ECG. Through interdisciplinary collaboration, we successfully deployed this system in real-world settings, revolutionizing emergent healthcare from in-hospital to out-of-hospital scenarios, thereby promptly aiding numerous patients experiencing myocardial infarction or severe dyskalemia.
The randomized controlled trial (RCT) is essential to determine safety and efficacy, enhance generalizability, inform health policy and practice, improve economic efficiency in healthcare, and guide future research and implementation in healthcare system. Due to the scarcity of real-world evaluation and underrepresentation in cardiology, the AI RCTs provide less clinically meaningful evidence for application in cardiovascular healthcare. We conducted pragmatic RCTs, identifying high-risk patients necessitating intensified management. Results demonstrated significantly reduced mortality rates, and expedited door-to-balloon times for STEMI patients in the intervention group compared to controls. Collectively, our AI-ECG system signifies a paradigm shift towards comprehensive and superior patient care within our healthcare institution. Further multi-centered validation of our AI-ECG system is warranted to substantiate its generalizability and assess its impact on healthcare expenditure.