Cardiac amyloidosis is the kind of diagnosis that tends to arrive late. The condition, caused by misfolded proteins building up inside heart tissue, mimics far more common forms of heart failure, and patients often see several specialists before anyone identifies the underlying cause. On April 8, 2026, the FDA cleared an AI tool built to shorten that delay: an algorithm from Massachusetts-based Anumana that reads a standard 12-lead ECG and flags signs of the disease, the first and only clearance of its kind for an ECG-based amyloidosis detector.
A Mayo Clinic Idea Turned Into a Company
Anumana’s ECG-AI model for cardiac amyloidosis began as research inside Mayo Clinic, where cardiologists had observed that the disease leaves subtle fingerprints on an ECG tracing that are easy for a human reader to overlook but detectable by a machine-learning model trained on enough examples. Anumana itself was co-founded by Mayo Clinic and nference, a data and AI company, specifically to commercialize algorithms developed from Mayo’s clinical research. The amyloidosis model is the latest in a string of ECG-AI products the company has pushed through the FDA pipeline, following an earlier clearance for detecting low ejection fraction.
To validate the amyloidosis algorithm beyond Mayo’s own patient population, Anumana ran a large independent, multicenter study spanning 25,525 patients across four U.S. health systems. In that study, the algorithm detected cardiac amyloidosis with 78.9% sensitivity and 91.2% specificity among adult patients who presented with signs, symptoms, or comorbidities associated with the condition, according to the company.
Why This Disease Is So Easy to Miss
Cardiac amyloidosis has historically been underdiagnosed for a combination of reasons: its symptoms, including shortness of breath and fatigue, overlap heavily with far more common conditions like standard heart failure or coronary artery disease, and confirming a diagnosis typically requires advanced imaging or a biopsy that physicians don’t order unless amyloidosis is already high on their list of suspects. That creates a diagnostic catch-22: doctors need a reason to suspect the disease before they’ll order the tests that confirm it, and the disease rarely announces itself clearly enough to prompt that suspicion.
Left unrecognized, amyloid deposits continue to stiffen the heart muscle, eventually leading to heart failure that can be fatal. Newer drug treatments for transthyretin amyloid cardiomyopathy, one of the two main subtypes, have made earlier diagnosis more clinically urgent than it used to be, since those therapies work best when started before the disease has caused extensive, irreversible damage. That treatment landscape is part of what has made an accessible screening signal, pulled from a test already performed millions of times a year, attractive to health systems.
The Case for Opportunistic Screening
Because ECGs are already part of routine cardiac workups, an algorithm that runs quietly in the background of existing tests offers a way to screen broadly without adding cost or new equipment. Supporters describe this as opportunistic screening: rather than ordering a dedicated amyloidosis test, the health system simply re-analyzes ECGs it was already collecting for other reasons, and flags the subset of patients who warrant a closer look with advanced imaging or blood biomarker testing.
Where the Caution Comes In
The flip side of a sensitivity-specificity profile like Anumana’s is that roughly one in five true amyloidosis cases will still be missed, and the specificity figure means a meaningful share of ECGs flagged as suspicious will turn out not to have the disease on further workup. Cardiologists reviewing the clearance have generally welcomed it as a screening aid rather than a diagnostic endpoint in itself, emphasizing that abnormal results still require confirmation through cardiac MRI, nuclear scintigraphy, or biopsy before a patient starts amyloidosis-specific treatment. There is also the recurring question of whether performance holds up as the tool moves from the four health systems in Anumana’s validation study to the full diversity of ECG machines, patient populations, and clinical settings found across the country.
Part of a Broader ECG-AI Push
Anumana’s clearance is one of several recent FDA authorizations for AI tools that extract disease signals from standard ECGs rather than relying on new imaging modalities, part of a broader industry bet that there is more diagnostic information embedded in the humble ECG than clinicians have historically been able to extract by eye. For health systems, the appeal is straightforward: the hardware is already installed, the test is already routine, and the marginal cost of running an additional algorithm against an ECG that would have been recorded anyway is close to zero.
What Happens Next
With FDA clearance secured, Anumana’s amyloidosis algorithm is now available for clinical use, and the company’s next challenge is adoption: persuading hospital cardiology departments to integrate the tool into their ECG review workflows and training clinicians on how to act on a positive flag. Reimbursement will likely shape how quickly that happens, since hospitals weigh new AI tools against the staff time needed to chase down confirmatory testing for every flagged patient. If early deployments show that flagged patients go on to receive amyloidosis-specific treatment earlier than they otherwise would have, that outcomes data is likely to become the strongest argument for wider rollout.
Photo: Myriams-Fotos / PIXABAY via Pixabay
