Advanced AI model uncovers high-risk heart failure phenotype in people with diabetes
An artificial intelligence (AI) model has demonstrated its potential to provide a “comprehensive characterisation” of diabetic cardiomyopathy, according to researchers from UT Southwestern Medical Center in Dallas. This breakthrough could significantly advance efforts to prevent heart failure in individuals with diabetes.
The study revealed that a machine learning model can effectively identify individuals living with diabetic cardiomyopathy—a disorder of the heart muscle in people with diabetes that can progress to heart failure.
According to the findings, the AI model is capable of detecting a high-risk phenotype associated with diabetic cardiomyopathy. This insight offers the possibility of earlier interventions, potentially averting the onset of heart failure.
Dr Ambarish Pandey, the study’s first author, explained:
“This research is noteworthy because it uses machine learning to provide a comprehensive characterisation of diabetic cardiomyopathy – a condition that has lacked a consensus definition – and identifies a high-risk phenotype that could guide more targeted heart failure prevention strategies in people with diabetes.
“Phenotypes are observable physical properties of individuals that give them specific biological traits.”
Study Design and Key Findings
The study involved an analysis of health data from 1,000 adults participating in the Atherosclerosis Risk in Communities cohort. All participants had diabetes but no prior history of cardiovascular disease.
Using this dataset, researchers assessed 25 echocardiographic parameters and cardiac biomarkers, ultimately identifying three distinct patient subgroups.
One of these subgroups, comprising approximately 27% of participants, was classified as the high-risk phenotype group. Members of this group exhibited elevated levels of NT-proBNP, a biomarker indicative of abnormal heart remodelling and cardiac stress.
Importantly, individuals within the high-risk phenotype group had a 12.1% higher likelihood of developing heart failure compared to those in the other two subgroups.
The findings suggest that between 16% and 29% of people with diabetes may exhibit this high-risk phenotype.
AI in Action: A Neural Network for Early Detection
To extend the impact of this research, the team developed a deep neural network classifier capable of identifying more cases of diabetic cardiomyopathy.
Dr Pandey elaborated:
“Clinically, this model could help target intensive preventive therapies, such as SGLT2 inhibitors, to patients most likely to benefit. “It may also help enrich clinical trials of heart failure prevention strategies in people with diabetes.”
Building on Previous Research
Dr Pandey highlighted that this study builds on earlier work examining the prevalence and prognostic implications of diabetic cardiomyopathy in adults living in the community.
“It extends those efforts by using machine learning to identify a more specific high-risk cardiomyopathy phenotype,” he said.
This innovative approach could help redefine how diabetic cardiomyopathy is diagnosed and managed, paving the way for more personalised and effective treatment strategies.
The full study can be accessed in the European Journal of Heart Failure.
