Research Summary

Nonrandomized Trial Finds AI Screening Detects ATTR-CM

Anthony Calabro, MA

Key Highlights

  • The ATTRACTnet AI model used electrocardiographic, echocardiographic, demographic, and diagnostic-code data to identify potential ATTR-CM.
  • Among 50 patients who completed further testing, 24 (48%) received an ATTR-CM diagnosis.
  • The testing positivity rate was more than 2.8 times higher than the 15.3% rate among historical controls.
  • Of the 24 patients diagnosed with ATTR-CM, 21 (88%) initiated treatment within 3 months.

An artificial intelligence (AI)–augmented screening program identified previously undiagnosed transthyretin amyloid cardiomyopathy (ATTR-CM) in nearly half of the patients who underwent subsequent diagnostic testing, according to a nonrandomized clinical trial published in JAMA Cardiology. The investigators developed the program to address the underdiagnosis of ATTR-CM despite the availability of expanding treatment options.

The researchers constructed an AI model called ATTRACTnet using electrocardiographic waveforms, echocardiographic measurements, demographic characteristics, and diagnosis codes for orthopedic manifestations of amyloidosis. The model was trained and validated at a large academic ATTR-CM referral center and externally validated at another academic site.

The investigators then evaluated the model in a single-system, multisite, single-arm, open-label trial. Patients were eligible if they had a left ventricular wall thickness of ≥12 mm and an ATTRACTnet score of ≥0.5. With approval from their treating physicians, eligible patients were notified and offered nuclear scintigraphy, monoclonal protein testing, and follow-up care. The primary outcome was an ATTR-CM diagnosis according to consensus criteria.

Study Findings

ATTRACTnet was developed in an internal test set of 799 patients with a mean age of 75.1 years; 64.7% were male and 35.3% were female. The model demonstrated an area under the receiver operating characteristic curve of 0.85 in the internal test set and 0.82 (95% CI, 0.81-0.83) in an external test set of 422 patients. Performance was similar among Hispanic, non-Hispanic Black, and non-Hispanic White patients.

During the trial, the AI model identified 1471 patients with scores of ≥0.5. Of these patients, 256 met the study eligibility criteria and 50 underwent amyloidosis testing after physician and patient approval. ATTR-CM was diagnosed in 24 of the 50 tested patients (48%), and 21 of those diagnosed (88%) initiated treatment within 3 months.

The ATTR-CM testing positivity rate was more than 2.8 times higher than the 15.3% rate among historical controls (95% CI, 13.1%-17.9%; P<.001). The program was also associated with an 18% relative increase in new ATTR-CM diagnoses compared with the previous year.

Clinical Implications

According to the study authors, AI-augmented screening may help identify patients with ATTR-CM who would otherwise be missed by usual care and potentially reduce diagnostic delays. The findings also demonstrated how an AI model could be incorporated into a clinical program that includes physician review, patient notification, confirmatory testing, and treatment initiation.

The researchers emphasized that prospective randomized trials are needed to determine whether AI-supported ATTR-CM detection improves clinical outcomes.

Expert Commentary

“AI-augmented screening may improve ATTR-CM detection and identify patients who are missed by usual care,” the researchers concluded.

Reference
Jain SS, Sun T, Pierson E, et al. Detecting transthyretin cardiac amyloidosis with artificial intelligence: a nonrandomized clinical trial. JAMA Cardiol. 2026;11(2):117-124. doi:10.1001/jamacardio.2025.459