AI Algorithm Accelerates Diagnosis and Enhances Care for High-Risk Heart Patients

Mount Sinai researchers have made significant strides in the identification and risk assessment of hypertrophic cardiomyopathy (HCM) through the calibration of an artificial intelligence (AI) algorithm. This innovative approach aims to enhance the accuracy of HCM detection, enabling healthcare professionals to provide timely and individualized care to those affected by this serious heart condition. The study, recently published in the journal NEJM AI on April 22, 2025, highlights the algorithm named Viz HCM, which had already received approval from the Food and Drug Administration for detecting HCM through electrocardiogram (ECG) readings.

The key advancement introduced in this study is the assignment of calibrated numeric probabilities to the algorithm’s assessments. Previously, the algorithm’s output included broad categorizations like “suspected HCM” or “high risk of HCM.” However, researchers at Mount Sinai have refined the process to give patients more concrete information—such as a specific likelihood of having HCM, which could, for example, indicate a 60% chance of the condition. Joshua Lampert, MD, who is the Director of Machine Learning at Mount Sinai Fuster Heart Hospital, emphasized the transformative potential of this detailed feedback for patients. It allows individuals who might not have been previously diagnosed with HCM to gain insights into their heart health, thereby facilitating early intervention and treatment.

The importance of such advancements cannot be overstated, especially considering that HCM affects approximately one in 200 people worldwide and remains one of the leading causes for heart transplantation. Many individuals with HCM are unaware they have the condition until symptoms arise, often when the disease has already progressed to a more severe stage. By integrating this AI tool into clinical workflows, doctors can identify high-risk individuals earlier, potentially preventing critical complications associated with HCM, such as sudden cardiac death—particularly affecting younger patients, who are often in the prime of their lives.

The research team under Lampert analyzed nearly 71,000 ECG readings collected from patients between March 2023 and January 2024. Out of these, the Viz HCM algorithm flagged 1,522 cases as showing potential signs of HCM. To validate the findings, researchers conducted an extensive review of patient records and imaging data to establish confirmed diagnoses of HCM. The results yielded promising conclusions: the calibrated AI model effectively provided an accurate correlation between its predicted probabilities of HCM and the actual incidence of the disease among patients.

Enhancing the interpretability of AI in healthcare has become a major focus in recent years, and this study serves as a prime example of how technology can be integrated into clinical practices to improve patient care. Clinicians can leverage this calibrated risk model to prioritize patients according to their individual levels of risk, ultimately streamlining clinical workflows. This change allows healthcare providers to offer more tailored guidance during consultations, transforming how patients experience the healthcare system.

Dr. Vivek Reddy, co-senior author and Director of Cardiac Arrhythmia Services for Mount Sinai Health System, remarked on the transformative potential of these developments in clinical practice. He noted that the utilization of novel algorithms like Viz HCM could significantly enhance patient triage and risk stratification processes. This methodological sophistication underscores the increasing importance of employing advanced AI tools not just for their performance but for their capacity to improve patient outcomes and align with existing clinical practices.

In addition to enhancing patient care through a clearer understanding of individual risks, the research also emphasizes the importance of pragmatic implementation in healthcare settings. Dr. Girish N. Nadkarni, another co-senior author and Chair of the Windreich Department of Artificial Intelligence and Human Health, highlighted that successful integration of AI into medical workflows hinges on its ability to support clinical decision-making while ensuring it aligns with how healthcare is delivered. This study exemplifies a responsible approach to the integration of AI, showcasing that a calibrated model can significantly aid clinicians in managing their patient populations more effectively.

Despite the promising outcomes of this study, the research team acknowledges that further exploration is required for the broader application of this AI calibration strategy across different health systems nationwide. The next phase of research will focus on expanding the use of the calibrated model to ensure its efficacy and adaptability across diverse clinical environments. The ultimate goal is to establish a standardized method for employing AI technology and machine learning algorithms to enhance the predictability and reliability of cardiac diagnoses.

The potential implications of this study extend beyond the realm of HCM, as they pave the way for implementing AI in addressing a wide variety of conditions. As advancements continue within the AI space, cardiologists and healthcare providers are encouraged to remain vigilant and informed about the technological innovations that can be utilized to enhance patient care.

As the healthcare landscape navigates the integration of innovative AI tools, it becomes increasingly critical for medical practitioners to embrace these changes for the benefit of their patients. With the ability to provide targeted risk assessments and improved clinical workflow efficiency, the use of calibrated AI models represents a significant leap forward in the medical field. Researchers and clinicians are hopeful that this will establish a new paradigm in cardiology and beyond, effectively revolutionizing how patients are diagnosed, treated, and monitored.

The Mount Sinai Health System, renowned for its commitment to exceptional cardiovascular care, holds a pivotal role in championing such initiatives. As collaborative efforts among researchers, healthcare professionals, and technology developers continue to flourish, the mounting evidence supporting the role of AI in enhancing accuracy, efficiency, and patient engagement in healthcare will undoubtedly continue to grow.

Keywords

Artificial intelligence, Hospitals, Human health, Heart disease, Risk factors, Machine tools, Cardiology, Cardiomyopathy, Electrocardiography.