Researchers from Weill Cornell Medicine, Cornell Tech, Cornell Ann S. Bowers College of Computing and Information Science, Columbia University Vagelos College of Physicians and Surgeons, and NewYork-Presbyterian announced on Mar. 23 that artificial intelligence techniques applied to cardiac ultrasound data can help identify patients with advanced heart failure. The findings were published in npj Digital Medicine.
This development could improve care for thousands of people who may not receive appropriate treatment due to the challenges involved in diagnosing advanced heart failure. Current diagnosis relies on cardiopulmonary exercise testing (CPET), which requires specialized equipment and staff available only at large medical centers.
The new AI method predicts peak oxygen consumption (peak VO2), a key CPET measure, using standard ultrasound images of the heart combined with electronic health records. This approach was developed through collaboration between informatics and AI experts led by Dr. Wang, clinicians including Dr. Nir Uriel from NewYork-Presbyterian, and Dr. Deborah Estrin from Cornell Tech.
"Initially we put together a group of more than 40 heart failure specialists and asked them to tell us where they thought AI could best be applied," said Dr. Uriel, who is also the Seymour, Paul and Gloria Milstein Professor of Cardiology at Columbia University Vagelos College of Physicians and Surgeons and an adjunct professor at Weill Cornell Medicine.
Dr. Estrin said: "The close interaction between clinicians and AI researchers on this project ended up driving the development of new AI techniques that would not have been explored otherwise... So, this was a case of medicine shaping the future of AI—not just AI shaping the future of medicine." The model processes several types of data including moving ultrasound images showing heart function as well as information from electronic health records.
The machine learning model was trained on deidentified data from 1,000 patients with heart failure treated at NewYork-Presbyterian/Columbia University Irving Medical Center before being tested on another group of 127 patients across three other campuses. The results showed about 85% accuracy in identifying high-risk patients—better than previously reported tools for predicting peak VO2 using artificial intelligence.
Looking ahead, clinical studies are planned to support regulatory approval by the U.S. Food and Drug Administration before routine use in healthcare settings can begin.
"If we can use this approach to identify many advanced heart failure patients who would not be identified otherwise, then this will change our clinical practice and significantly improve patient outcomes and quality of life," Dr. Uriel said.
