Machine learning spots hidden risk of functional decline after heart failure
For older adults with heart failure (HF), leaving the hospital alive does not always mean returning to an independent life. Loss of independence can reduce quality of life, increase caregiver burden and raise care needs. Yet most prognostic approaches focus on death or rehospitalization, making it difficult to identify patients likely to survive while losing independence.
Older adults who have experienced heart failure often find themselves facing the challenge of losing their independence, which can have a significant impact on their quality of life and increase the burden on caregivers. Despite the focus on death and rehospitalization in most prognostic approaches, researchers led by Kanji Yamada from Juntendo University in Japan have developed a machine learning model that can predict functional decline or death in these patients one year after discharge.
The study, published in the Journal of Cardiac Failure, analyzed data from 6,382 patients aged 65 years or older who were hospitalized for heart failure and remained independent in activities before admission. By using the J-Proof HF Registry, the researchers incorporated assessments performed by physical therapists at discharge, including gait speed, the Short Physical Performance Battery, handgrip strength, and the Barthel Index, as well as preadmission frailty measured by the Kihon Checklist.
The XGBoost model, which was compared to a modified KCL score, proved to be more effective in predicting functional decline and mortality at one year. The model identified four risk phenotypes, including a group of patients with low mortality but a high risk of functional decline despite an 89% probability of survival. This discovery highlights the importance of considering physical function alongside mortality risk when planning post-discharge care for patients with heart failure.
The researchers have developed a prototype web application that estimates individual risk at discharge and simulates the impact of changes in physical function on predicted risk. This tool could help clinicians and patients discuss rehabilitation needs earlier and make more efficient use of limited resources. Overall, the study demonstrates the potential of machine learning in identifying functional decline alongside mortality risk in older adults with heart failure, with the ultimate goal of preserving independence and quality of life.
Written by urgent.news from Medical Xpress's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
