AI Medical Tools May Allow for Simpler, More Accessible ECGs

​A new medical tool powered by artificial intelligence (AI) has simplified heart monitoring processes for medical professionals and patients. A Scripps Research initiative has demonstrated that AI-supported medical tools can help cardiologists diagnose heart attacks and related cardiovascular conditions using electrocardiogram technology that is simpler and more accessible to use.

In the past, physicians and surgeons have largely relied on 12-lead electrocardiograms (ECGs) to research and diagnose heart failure and abnormal heart rhythms. This poses several challenges due to the complex setup involved with 12-lead ECGs; each lead is a long wire attached to electrodes that must be placed at certain points on a patient's body, including their chest and limbs. The bulky, specialized equipment is hard to move and can only be operated by trained experts, limiting use among many clinicians and forcing patients to travel significant distances for 12-lead ECG analysis.

The introduction of AI support has allowed medical researchers to develop similar equipment that uses only three electrodes, making the process much simpler and more mobile. The initial report, first published in a 2024 issue of npj Digital Medicine, described the new setup's ability to produce reports on par with those yielded by 12-lead ECGs. The advent of such technology would provide countless patients with access to high-quality, time-sensitive data on their cardiovascular health. The new setup is also expected to decrease the cost of ECG technology.

The research team, headed by cardiologist Evan Muse, MD, PhD, centered the new equipment around an AI tool that draws on data from more than 600,000 12-lead ECGs collected from a diversity of patients. About half of the patient population used by the AI tool demonstrated healthy heart rhythms, while the remaining group lived with a wide range of cardiovascular conditions. Giorgio Quer, PhD, the director of artificial intelligence at Scripps Research Translational Institute, harnessed that data and began experimenting with two and three-electrode setups accurate enough to reflect a full 12-lead analysis.

Researchers initially wanted to limit the equipment to leads on patients' limbs, as these are the easiest electrodes for non-experts to work with. The team did not want to compromise accuracy or patient safety, however, and determined that much of the strongest data was collected when using a chest lead as well.

After devising the overall setup, the research team examined a set of 238 ECGs, 50 percent of which demonstrated signs of heart failure. Cardiologists were then asked to assess the ECGs, which provided researchers with two key successes. First, the cardiologists could not determine which ECGs came from the standard 12-lead setup and which ECGs had been partially constructed from AI data using three leads. Furthermore, cardiologists correctly identified heart failure symptoms in 81.4 percent of the AI-produced ECGs, just shy of the 84.6 percent on the original 12-lead ECGs.

While the AI tool has produced encouraging results, more studies using a wider array of patient groups will need to be completed. That said, the three-lead setup promises faster diagnoses and treatments for patients who would otherwise not have access to such technology. The technology is just one example of AI tools being used to screen for and diagnose different heart conditions, including atrial fibrillation.

Sanjiv Narayan

Shop