Yale Researchers Develop AI Model For Heart Condition Diagnosis

Yale School of Medicine researchers have developed an artificial intelligence (AI) model that can help diagnose heart rhythm and conduction disorders based on electrocardiograms (ECG) images.

Machine learning has become increasingly prominent in healthcare, as providers are increasingly using it to enhance diagnoses. Applying machine learning to ECGs can increase the likelihood of physicians locating patterns that often go unrecognized, researchers found.

The artificial intelligence-based model created by the researchers is based on data from more than 2 million ECGs from 1,506,112 patients. The model was able to diagnose one in six patients with a heart rhythm disorder.

“Our study suggests that image and signal models performed comparably for clinical labels on multiple datasets,” said Veer Sangha, a computer science major at Yale College, in the press release. “Our approach could expand the applications of artificial intelligence to clinical care targeting increasingly complex challenges.”

Researchers validated the model using international data sources. They published a study in Nature Communications, detailing the development and validation of the AI method.

The use of AI-based technology does come with various challenges, researchers noted.

“Current AI tools rely on raw electrocardiographic signals instead of stored images, which are far more common as ECGs are often printed and scanned as images. Also, designing many AI-based diagnostic tools is meant for individual clinical disorders, and therefore, may have limited utility in a clinical setting where multiple ECG abnormalities co-occur,” said Rohan Khera, MD, MS, in the press release. 

But the new AI approach is very advanced and can monitor several clinical diagnoses at once, similar to a human being, Khera added.

In recent years, AI-based technology has been used to predict, monitor, and treat of multiple conditions.

For example, an AI model created in January can assist physicians in predicting patients’ risk for a heart attack. The AI model works by analyzing coronary 18F-NaF uptake on PET and quantitative coronary plaque characteristics on CT angiography, indicating heart attack risk.

A separate study published in Cardiovascular Research similarly evaluated the risk for heart attack. Specifically, a machine-learning tool was used to combine the scoring of coronary artery calcium with non-contrast computed tomography, which are two factors that indicate heart attack risk.

Another study from December 2021 explained the development of an AI system that monitors the progress of chronic diseases as patients age. The model does this by evaluating metabolic and cardiovascular status, which indicates overall health.