Photo credit: www.sciencedaily.com
Researchers at Mount Sinai are utilizing advanced artificial intelligence (AI) to enhance the diagnosis of hypertrophic cardiomyopathy (HCM), a form of heart disease characterized by abnormal thickening of the heart muscle. The innovative AI algorithm, called Viz HCM, aims to identify patients at high risk for the condition more efficiently and accurately, providing critical information for discussions during medical consultations.
Previously, the Viz HCM algorithm received approval from the Food and Drug Administration for detecting HCM through electrocardiograms (ECGs). According to a study published on April 22 in the journal NEJM AI, this upgraded system now assigns specific numeric probabilities to its findings, allowing for clearer risk assessments. For instance, rather than simply labeling a patient as “suspected HCM” or “high risk,” the algorithm can articulate that a patient has a “60 percent chance of having HCM,” as explained by Joshua Lampert, MD, who leads the machine learning initiatives at Mount Sinai Fuster Heart Hospital.
This nuanced approach provides patients who might not have had a prior diagnosis of HCM with a clearer understanding of their individual risk levels. This knowledge can facilitate earlier and more tailored evaluations, as well as treatment approaches that significantly reduce the chances of serious complications like sudden cardiac death, which is particularly critical for younger patients.
Dr. Lampert expressed the importance of this advancement in integrating deep-learning algorithms into practical clinical settings. He noted that clinicians can streamline their workflows by using this sorting tool to highlight the highest-risk patients for priority attention. Furthermore, by providing patients with more detailed risk information through this calibrated model, healthcare professionals can enhance the quality of counseling and support offered. However, the applicability of this calibration strategy in broader medical contexts remains to be verified.
Hypertrophic cardiomyopathy affects approximately one in every 200 people globally and is a significant cause of heart transplantations. A major concern is that many individuals are unaware they have this serious condition until they exhibit symptoms, often indicating that the disease has progressed considerably.
In their study, the Mount Sinai team applied the Viz HCM algorithm to roughly 71,000 patients who underwent ECGs between March 7, 2023, and January 18, 2024. Out of this cohort, 1,522 patients received a positive alert for HCM. Follow-up reviews of their medical records and imaging data allowed researchers to confirm actual diagnoses.
The accuracy of the calibrated AI model was examined to see if its probability estimates aligned with the reality of patients’ HCM diagnoses. Findings indicated that the calibrated model reliably predicted the likelihood of disease presence in patients.
This AI integration in analyzing ECG data could significantly enable cardiologists to prioritize appointments for high-risk patients, potentially facilitating early treatment before the development of serious symptoms. Personalized discussions about risk—with specific probabilities rather than vague assessments—could engage patients more effectively in their healthcare, aiding in the prevention of HCM-related adverse events.
“This research provides crucial insights that could reshape our strategies for triaging and counseling patients,” stated co-senior author Vivek Reddy, MD, who oversees Cardiac Arrhythmia Services at Mount Sinai. He highlighted the potential of utilizing HCM as a model for operationalizing innovative tools in managing less common diseases through informed AI classifications.
Girish N. Nadkarni, MD, MPH, another co-senior author and chair of the Windreich Department of Artificial Intelligence and Human Health, emphasized the progressive implementation reflected in this study. He explained that recognizing the importance of merging high-performing algorithms with the realities of clinical practice is essential for enhancing patient outcomes. This study exemplifies how a calibrated AI model can assist clinicians in prioritizing patient care effectively.
The next phase involves extending the findings, including AI calibration for HCM, to various health systems across the nation, potentially transforming heart disease management practices.
Funding for this study was provided by Viz.ai. It is worth noting that Dr. Lampert is a compensated consultant for the same company.
Source
www.sciencedaily.com