A recent study published in Nature Communications Medicine introduces a non-invasive deep learning approach that analyzes electrocardiogram (ECG) signals to accurately predict a patient’s risk of developing heart failure. The model, called Cardiac Hemodynamic AI monitoring System (CHAIS), has shown promising results comparable to invasive procedures like right heart catheterization.
Researchers from MIT and Harvard Medical School used CHAIS to analyze ECG data from patients who were already scheduled for a catheterization procedure. The patch-based device provided accurate information on left atrial pressure, which is essential in characterizing cardiac function and optimizing treatment strategies for patients with heart failure.
CHAIS uses a single adhesive patch that can be worn outside of the hospital, eliminating the need for invasive procedures like right heart catheterization. This technology has the potential to aid in selecting patients who will most benefit from more invasive testing and serial monitoring of left atrial pressure in patients with heart disease.
The benefits of CHAIS extend beyond patient care, as it can help alleviate the burden on understaffed medical workforces by reducing hospital readmissions. The researchers hope to conclude their ongoing clinical trial soon, paving the way for data analysis and further development of this promising technology.
“This work is one step towards realizing our goal of providing equitable, state-of-the-art care to everyone, regardless of their socioeconomic status or background,” says Collin Stultz, senior author of the study.
Source: https://news.mit.edu/2025/can-deep-learning-transform-heart-failure-prevention-0210