AI flags heart-failure patients at risk of losing independence

For many older patients with heart failure, leaving the hospital alive is not the end of the ordeal. At Juntendo University in Japan, a team led by assistant professor Kanji Yamada wanted to know which of them would walk out and then slowly lose the ability to live on their own. They followed 6,382 patients aged 65 and over who had been hospitalized for heart failure, had received physical therapy and were independent before admission.

Most prognostic tools ask a single question: will the patient die, or be readmitted to hospital? Yamada's team built a model based on XGBoost, a machine-learning method that combines many decision rules, and fed it 77 candidate predictors. The strongest inputs were measured by physical therapists at discharge: maximum gait speed, the Short Physical Performance Battery (a set of balance, walking and standing tests), handgrip strength and the Barthel Index, a scale of daily-living tasks. Pre-admission frailty, measured with the Kihon Checklist, completed the picture. The model was evaluated using a nested leave-one-site-out method, which tests it on sites it did not learn from.

The results are clear on one point. Of the patients, 2,076 experienced functional decline and 928 died within a year, while 3,378 kept their independence. The ten-variable version of the model “significantly outperformed” the frailty-based scores. Gait speed and pre-admission frailty ranked among the most influential predictors, ahead of ejection fraction, the measure of how well the heart pumps. The model also sorted patients into four risk groups. The most striking was a group of 1,732 patients with a 89% chance of surviving the year but a 57% rate of functional decline. In other words, many of the people most at risk of losing their autonomy were not the ones most at risk of dying.

“Being able to live independently is just as important to many patients as how long they live,” Yamada says. “Instead of asking only whether a patient will survive, we need to ask whether they will be able to maintain the life they value after discharge.”

For clinicians, the practical promise is a shorter list of patients who need help. The team proposes that identifying the low-mortality, high-decline group could help teams prioritize more intensive outpatient or home-based rehabilitation after discharge. The researchers also built a prototype web application that estimates an individual's risk at discharge and simulates how faster gait speed might change it. Its authors are clear that it remains a prototype, and that implementation studies are still needed to show whether decisions guided by the model actually improve patient outcomes. No timeline for clinical use is given.