Predicting Hypertrophic Cardiomyopathy: AI Innovations That Could Change the Game
It’s estimated that as many as one in 500 people in the United States have hypertrophic cardiomyopathy (HCM), a heart disease where the heart muscle thickens (that’s the hypertrophy part) and reduces the amount of blood pumped through the body with each heartbeat. However, many people remain undiagnosed because they have no symptoms. “HCM is a challenging diagnosis because symptoms may be subtle—unusual fatigue or shortness of breath is often misattributed to other conditions,” says Kevin Shah, M.D., a board-certified cardiologist and the program director of heart failure outreach at MemorialCare Heart & Vascular Institute at Long Beach Medical Center in Long Beach, CA.
Imaging of the heart—ultrasound and cardiac magnetic resonance imaging (cardiac MRI)—is necessary to help confirm a diagnosis of HCM, Dr. Shah adds. “However, we don’t routinely do these tests unless we suspect a heart condition,” he says.
Furthermore, thickening of the heart muscle can be subtle and may not always be recognized as representing HCM, says Daniele Massera, M.D., the associate director of the hypertrophic cardiomyopathy program at NYU Langone in New York City: “The obstruction is often absent on imaging tests when the patient is at rest and needs to be provoked by maneuvers such as Valsalva (forcefully exhaling against a closed fist) or exercise.” If this is not routinely done, the obstruction can go undiagnosed for a long time and the patient won’t have access to appropriate treatment, he adds.
But with the advent of artificial intelligence in medicine, doctors may now be able to catch signs of a thickening heart wall sooner than ever before. Here, our experts give a closer look at what this could mean for the future of HCM care.
Diagonsis
AI in Early Detection and Diagnosis of HCM
When it comes to HCM, artificial intelligence may be able to improve rates of early detection and diagnosis, advance risk stratification for serious complications like sudden cardiac death, and develop personalized treatment strategies.
HCM is often an inherited, genetic condition caused by gene mutations. According to the American Heart Association (AHA), people with one parent with HCM have a 50% chance of having the gene change that causes the disease. “We do recommend genetic testing, since HCM is typically inherited,” says Dr. Shah. He adds that the HCM “gold standard” screening tool is the echocardiogram, as it directly shows the thickened heart muscle and measures obstruction. “We also often utilize cardiac MRI for more detailed imaging,” he says. “These tools are very effective when used, but the challenge is knowing when and who to screen. We can’t screen everyone, so we focus on family members of HCM patients and symptomatic individuals.”
By extracting much more information from that data doctors already use every day—such as the ECG, echocardiography, and cardiac MRI—AI is positioned to advance HCM diagnosis.
“Deep-learning ECG algorithms, for example, can detect subtle electrical signatures of hypertrophy that may precede overt morphological changes, while image-based models can automatically quantify wall thickness, ventricular function, and even myocardial fibrosis with high reproducibility,” says Giorgia Panichella, M.D., a cardiologist in the department of experimental and clinical medicine at the University of Florence in Italy.
An example of this is the Viz HCM algorithm calibrated by researchers from Mount Sinai Hospital in New York City. According to their study, this allows for interpretations like “you have about a 70% chance of having HCM” to give patients a better understanding of their individual disease risk.
“When these tools are integrated into the clinical workflow as decision-support systems, they can flag possible HCM earlier, reduce inter-observer variability, and help clinicians focus their attention on patients who need more in-depth evaluation,” says Dr. Panichella, who is part of the team behind SMASH-HCM, a project aiming to develop a cutting-edge digital-twin platform for personalized HCM stratification and improved disease management.