AI may accurately detect heart valve disease and predict cardiovascular risk

Advances in artificial intelligence have enabled the development and application of AI tools that could be effective in detecting heart valve disease and predicting cardiovascular disease risk, according to preliminary research in two studies to be conducted presented at the American Heart Association’s 2023 Scientific Sessions. held November 11-13 in Philadelphia.

“Computational methods to develop new predictors of health and disease – ‘artificial intelligence’ – are becoming increasingly sophisticated,” said Dan Roden, MD, FAHA, professor of medicine, pharmacology and biomedical informatics and senior vice president for personalized medicine at Vanderbilt University Medical Center and president of the Association’s Council on Genomic and Precision Medicine. “Both studies take a measurement that is easy to understand and easy to obtain and ask what that measurement predicts in the wider world.”

Real-world evaluation of an artificial intelligence digital stethoscope for detecting undiagnosed valvular heart disease in primary care

A study conducted at three different primary care clinics in the US compared the ability of a medical professional using a standard stethoscope to detect potential heart valve disease versus the ability of an artificial intelligence program using sound data from a digital stethoscope to do the same. to do.

Each study participant underwent a physical exam in which a primary care professional, a doctor or nurse, listened to their heart and lungs with a traditional stethoscope for unusual sounds or murmurs, plus an exam with a digital stethoscope that recorded heart sounds. All participants also received an echocardiogram at a follow-up appointment one to two weeks later to determine whether heart valve disease was present, although the results were not shared with the doctor or patient.

The analysis showed:

• The AI ​​method with the digital stethoscope detected 94.1% of cases of heart valve disease, compared to the standard stethoscope used by primary care professionals, which detected only 41.2% of cases.
• The AI ​​method identified 22 people with previously undiagnosed heart valve disease, and the professionals using the standard stethoscopes identified eight previously undiagnosed people with heart valve disease.

“The implications of an undiagnosed or late diagnosis of valvular heart disease are dire and impose significant costs on our health care system,” said lead author Moshe Rancier, MD, senior medical director of Mass General Brigham Community Physicians in Lawrence, Massachusetts.

“This study shows that healthcare professionals can more effectively and quickly screen patients for valvular heart disease using a digital stethoscope combined with high-quality AI that can detect heart murmurs associated with significant valvular heart disease.”

The study’s limitations include the small sample size of the study group, which precludes analyzes of differences between subgroups of participants (based on characteristics such as gender, race, ethnicity, and age).

Additionally, although the AI ​​method had greater sensitivity to sounds detected with the digital stethoscope, medical professionals using a standard stethoscope were able to be more specific in their diagnosis (95.5% vs. 84.5% for the AI ​​method), which can reduce the chance of false diagnoses. positives and/or additional tests or screenings for heart valve disease.

However, this study only evaluated the accuracy of the digital stethoscope compared to a traditional stethoscope. Rancier noted that researchers plan to evaluate patients’ six-month follow-up data to further assess clinical outcomes and additional diagnostic tests and treatments.

Study background and details:

• The study included 369 adults, all aged 50 and over, and 61% of participants identified as female.
• None of the participants had a previous diagnosis of valvular heart disease or a history of heart murmurs.
• The healthcare professionals who performed their routine examination on their patients were unaware of the AI ​​results or the echocardiogram results, making it a blinded study.
• Participants were enrolled from June 2021 through May 2023. Data collection and analysis is ongoing.
• The health care clinics where patients received care were located in Queens, New York, and Lawrence and Haverhill, Massachusetts.
• Study participants self-identified their race or ethnicity: 70% identified as white adults, 18% were Hispanic or Latino adults, 9% were black adults, and 2% identified as Asian adults; with 1% of participants identified as other.

“We saw here that the AI-based stethoscope did extremely well; it predicted almost 90% of the valve disease diagnoses that ended up being made. I see that as an emerging technology – taking an AI-based stethoscope and maybe making it combining it with other imaging modalities, such as an AI echocardiogram built into your stethoscope,” Roden said. “Using these new tools to detect the presence of valvular heart disease, as well as the extent of valvular disease and the extent of other types of heart disease, will likely help transform cardiovascular care.”

Deep learning-based retinal imaging for predicting cardiovascular disease in prediabetic and diabetic patients: a study using the UK Biobank

Using data from the UK Biobank, a second study by a different research group evaluated the effectiveness of using images of the retina at the back of the eye, which were analyzed by a deep-learning algorithm to reduce the risk of cardiovascular disease to predict, defined as heart attack, ischemic stroke, transient ischemic attack or death due to heart attack or stroke.

Deep learning is a method of artificial intelligence that trains computers to analyze multiple layers of data and gives computers the ability to ‘learn’ by developing their model independently of human intervention based on new information presented to it – a process that is being challenged due to the demands of both large amounts of computing power and data.

Previous research had successfully developed a deep learning algorithm to predict cardiovascular disease events by analyzing retinal images and coronary artery calcium scores.

Researchers used the deep-learning algorithm to categorize retinal images of 1,101 people with prediabetes or type 2 diabetes into groups with low risk, moderate risk and high risk for cardiovascular disease. They then measured the number of cardiovascular diseases among the participants over a median period of eleven years.

The analysis showed:

• 8.2% of participants in the low-risk group, 15.2% of participants in the moderate-risk group, and 18.5% of participants in the high-risk group had experienced cardiovascular disease by the end of the 11-year study period. and vascular diseases.
• After taking into account demographic and other potential risk factors for cardiovascular disease, such as age, gender, high blood pressure drug use, cholesterol medications, and smoking history, people in the moderate-risk group were 57% more likely to have a cardiovascular event than people in the low-risk group; and people with high-risk scores were 88% more likely to have a cardiovascular event compared to those in the low-risk group.

“These results demonstrate the potential of using AI analysis of retinal imaging as a tool for early detection of heart disease in high-risk groups such as people with prediabetes and type 2 diabetes,” said lead author Chan Joo Lee, MD, Ph. D., associate professor at Yonsei University in Seoul, Korea. “This could lead to early interventions and better management of these patient groups, ultimately reducing the incidence of heart disease-related complications.”

Study background and details:

• The UK Biobank is a large biomedical database and research resource containing health records of approximately 500,000 adults – enrolled from 2006 to 2010 – receiving care through the United Kingdom’s National Health Service. The researchers accessed the data in March 2023 and analyzed health records through June 2023.
• The participants were on average 59 years old; 45.5% were female and identified primarily as white race (85.5%).
• Of the 1,101 adults with prediabetes or type 2 diabetes, 550 people were in the low-risk group, 276 in the moderate-risk group and 275 in the high-risk group.
• At the end of the study period, 138 (12.5%) of the participants had experienced cardiovascular events: 45 were in the low-risk group; 42 were in the moderate risk group; and 51 belonged to the risk group.

The researchers tested the imaging’s ability to predict cardiovascular disease using a large data set of people, but this population was noted as having a predominantly white race, meaning the researchers’ findings may not apply to other populations . Additional follow-up studies are needed among people of different races and ethnicities.

“These systems learn from large data sets, and they only learn from the data we give them to learn from. For example, in the UK Biobank, 93% of participants are of European descent, so we have no idea whether the approaches coming from the UK Biobank are relevant or meaningful for people who are not of European descent,” said Roden.

“Another question is: Can the retinal scan predict coronary artery disease better than the pooled risk equations, or a polygenic risk score for coronary artery disease, or coronary calcium measurements? Those are all questions that need to be answered as we develop more tools to detect events like coronary artery disease. predict, we want to make sure we are using the right ones and the right combinations, rather than complicating care with alternative tools that have not been validated.”