Machine Learning Model Identifies Genetic Risk Factors in Heart Disease

Genetic factors in heart disease discovered by machine learning model

Cardiologists use magnetic resonance imaging (MRIs), which maps the structure of the heart, and electrocardiograms to get a better look inside. Since the two data types reveal different information about the heart and are usually studied separately, doctors can diagnose heart conditions.

In a paper that was published in Nature Communications by scientists at the Eric and Wendy Schmidt Centre of the Broad Institute of MIT & Harvard, they have developed a machine-learning approach to learn patterns from ECGs & MRIs simultaneously and predict heart characteristics based on these patterns. With further development, such a tool could help doctors detect and diagnose heart diseases from routine tests like ECGs.

Researchers also demonstrated that they can analyze ECG recordings which are inexpensive and easy to obtain, and produce MRI movies which are more expensive. Their method may even be used for finding new genetic markers that are associated with heart disease, which existing approaches that focus on individual data modalities could miss.