NEW prospective data suggest that artificial intelligence–enabled electrocardiogram (ECG) screening can be successfully implemented in real-world clinical settings to help identify patients with previously undiagnosed hypertrophic cardiomyopathy (HCM), a condition that often goes unrecognised until advanced disease or sudden cardiac events occur.
HCM is one of the most common inherited cardiac disorders, yet diagnosis is frequently delayed due to variable clinical presentation and limited sensitivity of traditional screening approaches. While AI-based ECG algorithms have shown promise in retrospective studies, evidence supporting their clinical implementation has been lacking. This multicentre prospective study aimed to evaluate how an AI-enabled ECG tool performs when deployed in routine care.
Real-World Implementation Across Health Systems
Researchers implemented an AI-based ECG algorithm, Viz HCM (Viz.ai), across five healthcare systems between January and December 2023. The tool automatically flagged 12-lead ECGs suggestive of HCM in adults without a prior diagnosis, alerting clinicians to review results and consider further evaluation.
Among nearly 146,000 ECGs screened, approximately 3% triggered an AI alert for suspected HCM. Clinicians reviewed just under 70% of flagged cases, demonstrating substantial engagement with the tool in busy clinical environments. From these reviewed alerts, 217 patients met eligibility criteria and were enrolled for follow-up.
Importantly, the cohort reflected broad racial and ethnic diversity, including Black, Asian, and Hispanic or Latino patients, supporting the generalisability of the approach across populations often underrepresented in cardiovascular research.
Accelerated Follow-Up and New Diagnoses
Most enrolled patients had a clinical indication for further investigation, prompting diagnostic imaging or specialist evaluation. The median time from ECG to confirmatory imaging suggestive of HCM was just over 7 days, highlighting the potential for AI-enabled alerts to shorten diagnostic delays.
Ultimately, 17 patients, nearly 8% of those enrolled, were newly diagnosed with HCM, with cases identified in both inpatient and outpatient settings. These findings suggest that AI-based ECG screening can uncover clinically meaningful disease that may otherwise remain undetected.
Refining Alerts Without Sacrificing Yield
During the study, researchers optimised the algorithm’s alert threshold to reduce unnecessary notifications. This adjustment nearly halved the proportion of ECGs flagged by the system without reducing the rate at which reviewed alerts led to patient enrolment. The result underscores the importance of balancing sensitivity with alert burden to support sustainable clinical adoption.
Implications for Clinical Practice
The study demonstrates that AI-enabled ECG screening for HCM can be integrated into diverse clinical workflows and meaningfully support case identification. While the absolute number of new diagnoses was modest, early detection of HCM has important implications for risk stratification, family screening, and prevention of adverse outcomes.
The authors note that further studies are needed to assess scalability, long-term outcomes, and how AI-based ECG screening compares with standard diagnostic pathways. Nonetheless, these findings represent an important step toward translating AI algorithms from proof-of-concept into routine cardiovascular care.
Reference
Love CJ et al. Clinical implementation of an AI-enabled ECG for hypertrophic cardiomyopathy detection. Heart. 2025;111(21):1029-1035.
