A recent study published in npj Digital Medicine suggests that a deep learning model using transthoracic echocardiograms (TTEs) can help predict patients with active or hidden atrial fibrillation (AF). The study conducted by Neal Yuan, M.D., from the University of California in San Francisco, and colleagues, created a two-stage deep learning algorithm that utilized a video-based convolutional neural network model to differentiate between TTEs in sinus rhythm or AF. The model was trained on 111,319 TTE videos.
The results of the study showed that the deep learning model was able to accurately differentiate TTEs in AF from those in sinus rhythm in a test cohort, with high accuracy levels (AUC 0.96, AUPRC 0.91). Furthermore, the model was also successful in predicting concurrent paroxysmal AF among TTEs in sinus rhythm in both the test and external cohorts. The model’s performance was consistent across various patient demographics and clinical factors, outperforming traditional methods of risk assessment.
The authors of the study believe that utilizing deep learning by TTE could provide new opportunities for identifying patients who may benefit from more intensive monitoring for atrial fibrillation. This research highlights the potential for technology to enhance patient care and screening processes in the future.
For more information on the study, you can access the full text here.
