Why it’s Notable:
A major finding of the Echo-Net trial was that the cardiologists made less corrections to the AI-generated assessment reports (16.8%) compared to the human sonographers’ reports (27.2%), demonstrating that the AI-based assessment reports were more indistinguishable from the cardiologists’ interpretations.
The AI-based assessment reports’ readings were also closer to the correct ejection fraction, at 2.8 percentage points away from the correct reading compared to the difference of 3.8 points for the sonographers’ assessment reports.
The trial conducted by the Smidt Heart Institute and the Division of Artificial Intelligence in Medicine at Cedars-Sinai is the first large prospective randomized controlled trial completed of an AI algorithm in cardiology.
Industry Implications:
Previous digital solutions utilizing AI-assisted algorithms to measure heart function include EchoNous, which showed clinical potential in a validation study in January 2022 for its handheld ultrasound device with an AI algorithm for automated quantification of left ventricular ejection fraction in real time.
Ultromic’s EchoGo Core is another solution utilizing AI and is undergoing clinical validation in a prospective randomized controlled trial within the UK's NHS, evaluating the use of AI for stress echocardiography.
The FDA has already approved a number of AI based algorithms for analysis support and detection in cardiology. The superior endpoints achieved by AI in the Echo-Net trial provide evidence for future improved diagnostic capabilities led by AI technologies in cardiology.
The diagnostic accuracy of AI algorithms are not only improving, but are proving more superior compared with human-based assessments. Artificial intelligence won't replace clinicians, but clinicians who utilize AI will have advantages over clinicians who don’t.
Read the full article here