MIT AI Predicts Heart Failure a Year in Advance

Imagine if doctors could look a year into the future and know exactly which heart failure patients need urgent care and which ones are doing fine.

That's exactly what researchers at MIT, Mass General Brigham, and Harvard Medical School just made possible. They developed an AI model called PULSE-HF that predicts whether a heart patient's condition will worsen within the next 12 months, using nothing more than a simple heart rhythm test.

Heart failure affects millions of people worldwide, weakening the heart muscle so it can't pump blood effectively. About half of people diagnosed with the condition die within five years. Knowing who's at highest risk could be lifesaving.

The AI focuses on something called left ventricular ejection fraction, which measures how much blood the heart pumps out with each beat. A healthy heart pumps 50 to 70 percent. When that number drops below 40 percent, patients face the most severe form of heart failure.

"Understanding how a patient will fare after hospitalization is really important in allocating finite resources," says Teya Bergamaschi, an MIT PhD student who helped create the model.

The team tested PULSE-HF on three different groups of patients from Massachusetts General Hospital, Brigham and Women's Hospital, and a public database. The AI achieved accuracy scores between 0.87 and 0.91 on a scale where 1 is perfect and 0.5 is random guessing.

Here's the really exciting part: The researchers also built a version that works with just one electrode on the body instead of the standard 12. That single-lead version performed just as well as the comprehensive test, making it perfect for rural doctors' offices that don't have full cardiac equipment.

The Ripple Effect

This breakthrough could transform how we manage heart failure across the globe. High-risk patients identified by PULSE-HF can get immediate follow-up care and closer monitoring. Meanwhile, lower-risk patients can skip unnecessary hospital visits and invasive testing, freeing up medical resources for those who need them most.

The tool works especially well in areas with limited access to cardiac specialists. A simple electrode and the AI model could help a rural family doctor provide the same quality predictions as a major city hospital with full cardiac imaging suites.

The project took years to complete, with the team facing challenges like cleaning messy medical data and dealing with imperfect readings from restless patients. But their persistence paid off with a tool that doesn't just detect existing heart problems but actually forecasts future ones.

No other method currently exists for predicting future heart decline in this way, making PULSE-HF a genuine first in the field.