Reviewed by Lexie CornerApr 23 2025
Researchers at Mount Sinai have calibrated an artificial intelligence (AI) algorithm to more efficiently and accurately identify patients with hypertrophic cardiomyopathy (HCM), a form of heart disease, and flag them as high-risk for additional attention during doctor’s appointments.
Prospective Validation of Calibrated Probability Scores and Impact of Probability Score Sorting. Image Credit: Reproduced with permission from NEJM AI, Lampert, 2025. Copyright 2025 Massachusetts Medical Society
The Food and Drug Administration (FDA) had previously authorized the algorithm, called Viz HCM, for detecting HCM on an electrocardiogram (ECG). The Mount Sinai study enhances the algorithm by assigning numerical probabilities to its results.
According to the Mount Sinai study, the algorithm may have previously flagged results as "suspected HCM" or "high risk of HCM," but the interpretations of these results varied.
You have about a 60 percent chance of having HCM.
Joshua Lampert, MD, Study Corresponding Author and Director, Machine Learning, Mount Sinai Fuster Heart Hospital
Patients who had not previously been diagnosed with HCM may gain a clearer understanding of their personal risk for the condition, enabling quicker and more tailored evaluations and treatments to potentially prevent complications, such as sudden cardiac death, especially in younger individuals.
Lampert, an Assistant Professor of Medicine (Cardiology and Data-Driven and Digital Medicine) at the Icahn School of Medicine at Mount Sinai, added, “This is an important step forward in translating novel deep-learning algorithms into clinical practice by providing clinicians and patients with more meaningful information. Clinicians can improve their clinical workflows by ensuring the highest-risk patients are identified at the top of their clinical work list using a sorting tool.”
He continued, “Patients can be better counseled by receiving more individualized information through model calibration, which improves interpretability of model classification scores. Whether this local model calibration strategy is universally applicable to other settings remains to be demonstrated. This can transform clinical practice because the approach provides meaningful information in a clinically pragmatic fashion to facilitate patient care.”
HCM is a leading cause of heart transplants and affects approximately one in 200 people worldwide. However, symptoms often appear after the disease has progressed, and many individuals remain unaware that they have the condition.
Mount Sinai researchers tested the Viz HCM algorithm on around 71,000 individuals who underwent electrocardiograms between March 7, 2023, and January 18, 2024. The system identified 1,522 individuals as having a positive HCM alert. To confirm whether these individuals had a verified diagnosis of HCM, researchers reviewed medical records and imaging data.
After validating the diagnoses, researchers calibrated the AI tool's model to assess whether the calibrated probability of having HCM aligned with the actual likelihood of patients having the condition. The calibrated model successfully provided a reliable assessment of a patient's probability of having HCM.
By using the model to review patients' ECG results, cardiologists could prioritize high-risk patients and schedule appointments and treatments earlier, before symptoms develop or worsen.
Rather than merely indicating that the AI model identified a patient, doctors would be able to communicate the specific risk to each patient. This approach could help prevent negative outcomes associated with HCM, such as sudden death or symptoms caused by thickening of the heart muscle that impedes blood flow.
This study provides much-needed granularity to help rethink how we triage, risk-stratify, and counsel patients. In an era of augmented intelligence, we must grow to incorporate novel sophistication in our approach to patient care. Using hypertrophic cardiomyopathy as an illustrative use case, we show how we can pragmatically operationalize novel tools even in the setting of less common diseases by sorting AI classifications to triage patients.
Vivek Reddy, MD, Study Co-Senior Author and Director, Cardiac Arrhythmia Services, Mount Sinai Health System
Study co-senior author Girish N. Nadkarni, MD, MPH, Chair of the Windreich Department of Artificial Intelligence and Human Health, Director of the Hasso Plattner Institute for Digital Health, and Irene and Dr. Arthur M. Fishberg Professor of Medicine at the Icahn School of Medicine at Mount Sinai, stated, “This study reflects pragmatic implementation science at its best, demonstrating how we can responsibly and thoughtfully integrate advanced AI tools into real-world clinical workflows. It is not just about building a high-performing algorithm—it’s about making sure it supports clinical decision-making in a way that improves patient outcomes and aligns with how care is actually delivered. This work shows how a calibrated model can help clinicians prioritize the right patients at the right time, and in doing so, help realize the full potential of AI in medicine.”
The next step is to expand this research and AI calibration for HCM to other health systems across the country.
Viz.ai funded this research. Dr. Lampert is a paid consultant for Viz.ai.