Points
- The main goal was to determine whether the artificial intelligence model is better compared to sonographers for echocardiograms.
- The results of this study also indicated that artificial intelligence (AI) saves time for both sonographers and cardiologists.
- Cardiologists were not able to differentiate between the assessments made by Sonographers and also based on AI.
Summary
Cardiologists presented during the 53rd Annual European Society of Cardiology Congress held online (26th to 29th August), Barcelona, Spain, that AI through a deep learning model is suitable for echocardiography assessment, and considerable changes were in final reports made by a cardiologist. According to David Ouyang (MD) of Cedars-Sinai Smidt Heart Institute, Los Angeles, deep learning is useful and gives more precision in phenotyping to find unknown previous risk factors.
Dr. Ouyang carried out the single-center EchoNet-RCT trial to evaluate the Artificial intelligence deep learning model for cardiac beat function. The trial includes 25 sonographers having experience of 14.1 years and 10 cardiologists with 12.7 years of expertise. Simpson’s biplane or Single plane methods were used for transthoracic echocardiography of 3495 adults. The echocardiography results were evaluated by a sonographer which was 1755, or Artificial intelligence, which were1740. Further, these results were checked by a Cardiologist.
The primary outcome was that substantial change as > 5% left ventricular ejection fraction (LVEF) resulted from the initial to the final assessment by cardiologists. Second, time for sonographer, cardiologist, and initial assessment made by cardiologist and historical cardiologist. The study trials for echocardiography included males with average ages were 66% and 57%. This group of patients has hypertension (38%), coronary artery disease (31%), diabetes (25%), chronic kidney disease (25%), atrial fibrillation (25%), and prior stroke (13%). The left ventricular ejection fraction was 58.1%; meanwhile, cardiologists were correct for only 32% time. Changes were found in the assessment made by a cardiologist as 50.1% for AI as compared to 54.5% for the sonographer group. Assessment times were found to be lower in artificial intelligence as compared to echocardiography by sonographer and cardiologist. Assessment of LVEF through artificial intelligence was used to assess LVEF and found artificial intelligence was not inferior but superior to sonographer assessment.
Link to the article: https://www.practiceupdate.com/content/esc-2022-artificial-intelligence-beats-out-sonographers-for-echocardiogram-initial-assessment/141447/6/2/2
References Ouyang, D. (2022, September 11). ESC 2022: Artificial Intelligence Beats Out Sonographers for Echocardiogram Initial Assessment. PracticeUpdate; PracticeUpdate. https://www.practiceupdate.com/content/esc-2022-artificial-intelligence-beats-out-sonographers-for-echocardiogram-initial-assessment/141447
