Estimated reading time: 4 minutes
A photo of a routine electrocardiogram may contain enough information for artificial intelligence to flag patients at risk for a potentially deadly heart disease that often goes undiagnosed until serious complications develop.
Researchers at Yale School of Medicine developed an AI platform that analyzes images of electrocardiograms, or ECGs, to identify patients who may have transthyretin amyloid cardiomyopathy, a condition caused by the buildup of misfolded proteins in the heart.
The technology, described in JAMA, was tested across eight separate patient cohorts in the United States and Europe and successfully identified people with the disease subtype. The U.S. Food and Drug Administration is now reviewing it.
For nurses, the research is notable because ECGs are already widely used across healthcare settings, while amyloid cardiomyopathy can be difficult to recognize. Its symptoms can overlap with other cardiovascular conditions, so patients may not be diagnosed until the disease has progressed to heart failure or other serious complications.
“We we able to leverage just an image from a very simple ECG into a screening test for cardiac amyloid,” said Rohan Khera, MD, director of the Cardiovascular Data Science Lab at Yale School of Medicine and the study’s principal investigator. “For a disease that’s massively underdiagnosed, identifying those at risk is very critical.”
Why Amyloid Cardiomyopathy Can Be Missed
Amyloid cardiomyopathy occurs when abnormal proteins form deposits in the heart, displacing healthy cardiac muscle.
The Yale study focused on transthyretin amyloid cardiomyopathy. The liver produces transthyretin, which can become misfolded because of genetics or age. As the protein accumulates in the heart, it can make the organ stiffer and interfere with the electrical system that regulates heart rhythm.
Without treatment, the disease can eventually lead to heart failure or death. According to the researchers, average life expectancy for untreated transthyretin amyloid cardiomyopathy is five years.
“It’s as aggressive as some of the most aggressive cancers,” Khera said.
Because symptoms can resemble other cardiovascular conditions, cases may not be recognized until complications develop. To address that gap, researchers wanted to determine whether a test already routinely used in cardiac care could reveal patterns clinicians might otherwise miss.
Turning an ECG Image Into a Screening Tool
ECGs measure the heart’s electrical activity using electrodes and are among the most commonly performed cardiac tests, with hundreds of millions performed worldwide each year. They are not currently used to screen patients for amyloid cardiomyopathy.
Researchers in Yale’s Cardiovascular Data Science Lab first trained an AI model to interpret ECG results using data from thousands of de-identified patients. They then used data from several hundred patients already diagnosed with amyloid cardiomyopathy to teach the model to recognize ECG patterns associated with the disease.
The platform can analyze an ECG image rather than requiring the original digital ECG data.
“You can take a photo of an ECG like you would take a photo of a bank check,” Khera said. “Our model can pick up from that photo whether the ECG came from somebody at risk for cardiac amyloid or not.”
Testing across eight distinct cohorts in the United States and Europe showed the model could identify patients with transthyretin amyloid cardiomyopathy.
“Our tool can really narrow down the funnel for who should be further evaluated for cardiac amyloid,” said Philip Croon, MMed, associate research scientist at Yale School of Medicine and the study’s first author.
The technology is intended to screen at-risk patients who may warrant further evaluation, not to replace clinical assessment.
What the Research Could Mean for Nurses
For nurses working in cardiac care, emergency departments, primary care, and other settings where ECGs are routinely performed, the research shows how AI could potentially extract additional information from a familiar clinical test.
Instead of requiring a new screening procedure, the platform analyzes an existing ECG image for patterns associated to a disease clinicians may not initially suspect. Earlier identification could help direct at-risk patients toward additional evaluation before extensive cardiac damage develops.
The technology is not yet cleared for routine clinical use. The research team received FDA Breakthrough Device Designation while developing the platform, and the tool is currently undergoing FDA review. The designation provides an expedited review process for certain medical devices.
Researchers are also studying how this type of AI screening performs in clinical settings. In the TRACE-AI Network Study, researchers are evaluating multimodal AI tools at 13 health centers across the United States. The study will examine their ability to detect transthyretin amyloid cardiomyopathy on a larger scale.
For nurses, those next steps will help show how AI-generated risk findings could ultimately fit into patient evaluation and care if the technology moves into clinical practice.
“We’ve been able to solve a critical bottleneck in deploying therapies by identifying more at-risk people,” Khera said. “This is one of our biggest accomplishments—making care accessible by finding people who need treatment the most.”
