Researchers at UT Southwestern Medical Center have created an AI tool that can find patients with a heart problem called diabetic cardiomyopathy. This condition causes changes in the heart’s structure and function, putting people with diabetes at a higher risk of heart failure. The study, published in the European Journal of Heart Failure, introduces a new way to detect these high-risk patients early, which can help doctors take steps to prevent heart failure.
“This research is important because it uses machine learning to better understand diabetic cardiomyopathy, a condition that is not well-defined. It also helps identify patients who are at high risk and could benefit from targeted heart failure prevention,” said Dr. Ambarish Pandey, a senior author and Associate Professor at UT Southwestern.
The study used data from over 1,000 people with diabetes but no known heart disease, gathered from the Atherosclerosis Risk in Communities project. Researchers analyzed 25 heart-related measures and identified three groups of patients. One of these groups, which made up 27% of the participants, showed a higher risk of heart failure. People in this group had high levels of a heart stress marker called NT-proBNP, and signs of abnormal heart changes, such as increased heart mass and poor heart function. Over five years, 12.1% of this group developed heart failure, a much higher rate than the other groups.
Based on this, the researchers developed a deep learning model to detect diabetic cardiomyopathy. When tested on other patient groups, the model found that 16% to 29% of diabetes patients had the high-risk condition. These patients consistently showed a higher chance of heart failure.
“This tool could help doctors focus on preventive treatments, like SGLT2 inhibitors, for patients who need them most,” said Dr. Pandey, referring to medicines used to treat Type 2 diabetes. “It could also improve clinical trials for preventing heart failure in diabetes patients.”
The study builds on previous research into diabetic cardiomyopathy, a condition that’s hard to diagnose early because it often doesn’t show symptoms. Machine learning is providing a better way to find high-risk patients compared to traditional methods.
By identifying people at risk of heart failure earlier, this model could lead to better treatments and improved patient outcomes. The research also aligns with UTSW’s goal to use data science and cardiovascular research to advance patient care.
Other researchers involved in the study were Dr. DuWayne Willett and Dr. Muhammad Shariq Usman from UT Southwestern. The research was funded by Applied Therapeutics.
Source: newswise
