Innovative AI Tool Enhances Detection of High-Risk Heart Patients for Faster, Personalized Care

Researchers from Mount Sinai have advanced the use of artificial intelligence (AI) in cardiology by refining an algorithm designed to identify patients with hypertrophic cardiomyopathy (HCM), a serious heart condition. The AI tool, known as Viz HCM, has been calibrated to provide more precise risk assessments by assigning numerical probabilities to its findings, allowing for a better understanding of individual patient risk levels.

Previously approved by the Food and Drug Administration for detecting HCM on electrocardiograms (ECGs), Viz HCM now offers calibrated risk scores, such as indicating a 60% chance of having HCM. This progress helps clinicians quickly pinpoint high-risk patients who might have otherwise gone undiagnosed until symptoms appeared, often at an advanced stage. Early detection is crucial for preventing severe outcomes such as sudden cardiac death.

The study, published in NEJM AI, analyzed nearly 71,000 patients' ECG data, where the algorithm flagged 1,522 individuals as potentially having HCM. After review and confirmation of diagnoses, the researchers applied model calibration to verify that the AI's probability scores aligned with actual disease presence. This approach empowers healthcare providers to prioritize those at highest risk for expedited evaluation and treatment.

Dr. Joshua Lampert, the study's lead, highlighted that the calibrated model provides meaningful and interpretable information, aiding both clinicians and patients. It enhances workflow efficiency by helping doctors focus on patients who need immediate attention, and enables more personalized patient counseling based on individual risk profiles.

The significance of this AI-based approach extends beyond HCM; it demonstrates a viable pathway for integrating deep learning tools into real-world clinical settings. By effectively triaging patients, particularly those with less common but serious conditions like HCM, healthcare systems can improve outcomes and optimize resource allocation.

HCM affects approximately 1 in 200 people worldwide and is a leading cause of heart transplantation. Many individuals remain undiagnosed until symptoms emerge or the disease progresses significantly. The Mount Sinai team tested Viz HCM on patients' ECGs collected over nearly a year, successfully calibrating the AI scores to closely match actual diagnoses. This calibration allows clinicians to better stratify patient risk, potentially leading to earlier interventions and the prevention of adverse events such as sudden cardiac death.

The researchers emphasize that the next step involves expanding this technology across various health systems to validate its effectiveness further. Overall, this innovative use of AI exemplifies how advanced algorithms can support clinical decisions, ultimately improving patient care and outcomes.