A screening tool for amyloid cardiomyopathy, a heart condition that often escapes diagnosis, has been developed at Yale. It reads pictures of ECG tracings rather than raw signal files, and it was built to run on a smartphone. The research appears in JAMA.
The condition begins when proteins that misfold accumulate in heart muscle. Early stages are easy to miss, and many people are diagnosed only once a serious complication has already appeared.
Reading a picture of a tracing
The platform sidesteps the usual requirement that a tracing be available as clean signal data. It works from the image, the format a clinician or patient is most likely to have on hand. That choice is what makes a phone a plausible screening device, since no specialized pipeline or vendor-specific export is needed.
The study’s principal investigator, Rohan Khera, who directs the Cardiovascular Data Science Lab at Yale School of Medicine, said the goal is to flag at-risk patients long before the disease announces itself.
Khera’s lab has a track record in ECG-based AI, and the group frames the tool as a way to move diagnosis earlier in a condition where damage accrues quietly. How the model performs across different ECG machines, image qualities and patient populations will decide whether screening translates into fewer late diagnoses.
