[Skip to Navigation]
Invited Commentary
August 4, 2021

Leveraging Large Clinical Data Sets for Artificial Intelligence in Medicine

Author Affiliations
  • 1Department of Cardiology, Cedars-Sinai Medical Center, Los Angeles, California
  • 2Division of Artificial Intelligence in Medicine, Department of Medicine, Cedars-Sinai Medical Center, Los Angeles, California
JAMA Cardiol. 2021;6(11):1296-1297. doi:10.1001/jamacardio.2021.2878

The last decade has seen extraordinary progress in the development and refinement of artificial intelligence (AI)–enabled tools for clinical applications.1,2 In cardiovascular medicine, deep learning algorithms using data from echocardiography and electrocardiography (ECG) have demonstrated superhuman performance in tasks typically done by physicians1,3 and identified subtle phenotypes invisible to a clinician’s routine review.4,5 Progress has been particularly quick in AI interpretation of ECGs, where the input data are well structured and the medical infrastructure for data acquisition and storage is relatively standardized across health systems. Already, there have been randomized clinical trials4 showing the effects of ECG-based AI algorithms on patient care. Compared with other forms of medical imaging, ECG waveforms routinely undergo signal processing and standardization that makes generalization across systems and devices easier. The ongoing development of AI systems for interpretation of ECGs is a promising avenue for extracting additional value from this readily available medical test.

Add or change institution