Malaysia
Singapore
Philippines
Stay connected with BizTech Community—follow us on Instagram and Facebook for the latest news and reviews delivered straight to you.
A groundbreaking artificial intelligence (AI) tool, PanEcho, may revolutionize echocardiogram interpretation by providing faster and highly accurate results, according to late-breaking research presented today at the American Heart Association’s Scientific Sessions 2024.
A Leap in Echocardiography
PanEcho, developed by the Cardiovascular Data Science (CarDS) Lab at Yale School of Medicine, is a comprehensive AI system capable of interpreting echocardiograms — videos that capture heart structure and function — with exceptional precision. Unlike previous AI applications that focused on single heart views or specific conditions, PanEcho analyzes multiple views and generates detailed reports across all major diagnostic categories.
“This tool has the potential to simplify echocardiogram screenings, especially in settings where expert readers are unavailable, allowing for quicker identification of urgent conditions,” said Gregory Holste, M.S.E., a researcher at the CarDS Lab, who presented the findings.
Key Findings
PanEcho’s performance was measured across 18 diagnostic tasks, achieving an average area under the receiver operating characteristic curve (AUC) of 0.91, where 1 represents perfect accuracy. Highlights include:
Left Ventricle Analysis
- Increased size detection: 0.95 AUC
- Systolic dysfunction identification: 0.98 AUC
- Left ventricular hypertrophy detection: 0.91 AUC
Valvular Disease Diagnosis
- Severe aortic stenosis: 0.99 AUC
- Mitral stenosis: 0.96 AUC
- Moderate or greater aortic regurgitation: 0.93 AUC
- Moderate or greater mitral regurgitation: 0.96 AUC
The AI system was also evaluated for its accuracy in estimating continuous parameters such as left ventricle ejection fraction and wall thickness. It demonstrated a mean absolute error of just 4.4% for ejection fraction estimates, critical for assessing heart function.
Implications for Real-World Use
PanEcho, tested on over 1.2 million echocardiogram videos collected from nearly 34,000 patients, offers a promising solution to reduce diagnostic bottlenecks in clinical settings. The system’s potential extends to:
- Emergency rooms and small clinics, where portable echocardiogram machines are frequently used.
- Underserved areas, lacking immediate access to cardiology experts.
“This represents a significant advancement in AI-driven healthcare,” Holste noted. “Our goal is to integrate PanEcho into real-world patient care environments to maximize its impact.”
Challenges and Future Directions
Despite its impressive performance, PanEcho has only undergone retrospective validation. Researchers aim to conduct prospective studies in clinical settings and further test the tool’s compatibility with portable echocardiogram devices.
Additionally, PanEcho’s ability to support multi-view, multi-task echocardiography interpretation could inspire broader adoption of flexible AI systems in medical diagnostics.
Demographic Considerations
The study population included a majority of adults aged 67, with 80% identifying as white, 14.2% as Black, and 1.8% as Asian. About 52% were men. While the results are promising, ensuring diverse representation in future research will be critical to validate PanEcho’s utility across different populations.
Promising the Future of Cardiology
“The progress made here exemplifies how AI can transform diagnostic accuracy and efficiency,” said Dr. Rohan Khera, head of the CarDS Lab and principal investigator of the study.
PanEcho’s public release is expected to catalyze further innovation, encouraging collaboration among researchers to refine multi-task AI tools for medical imaging.
For now, PanEcho’s creators are optimistic about its role in improving access to timely cardiac care, particularly in high-need areas, and its potential to set a new standard in echocardiography interpretation.