A research team from Juntendo University in Japan has developed a new model to predict mortality risk in elderly patients with heart failure. The model incorporates physical function metrics, which improves the accuracy of risk prediction compared to existing models that focus mainly on cardiac-specific clinical variables.
Current risk assessment tools for heart failure, such as AHEAD and BIOSTAT compact, were designed for European and North American populations. Studies have shown these models tend to underestimate the risk of death among older East Asian patients. To address this gap, researchers led by Professor Tetsuya Takahashi, Assistant Professor Kanji Yamada, and Associate Professor Nobuyuki Kagiyama analyzed data from the nationwide J-Proof HF registry. This registry tracks elderly heart failure patients treated at 96 institutions across Japan.
The team used machine learning algorithms on data from 9,700 patients who were discharged between December 2020 and March 2022. They trained an eXtreme Gradient Boosting (Full XGBoost) algorithm to estimate one-year mortality risk after hospital treatment for heart failure.
A second version of the model was created using only the top 20 most important variables identified by the first model. Of these variables, seven related to physical function or other non-cardiac factors. Dr. Yamada commented on the findings: "These models rely primarily on cardiac-specific and biomedical variables, often underestimating the impact of non-cardiac factors such as physical function, frailty, and nutritional status, which are critical determinants of prognosis in older adults and, unlike fixed factors such as age, may represent modifiable targets through rehabilitation and supportive care."
Dr. Yamada further noted: "The prominence of the BI [Barthel Index] and SPPB [Short Physical Performance Battery] in our analysis is clinically coherent," adding that "Unlike subjective activities of daily living assessments included in some scores, performance-based assessments, such as the BI and SPPB, offer greater reproducibility and capture functional limitations more directly."
Both versions of the XGBoost model showed similar accuracy in predicting death within a year after discharge. The Top-20 XGBoost model also classified patient risk more effectively than previous tools like AHEAD and BIOSTAT compact when applied to Japanese patients.
This approach could allow healthcare providers to identify elderly heart failure patients who would benefit from closer monitoring or tailored post-discharge care instead of relying on a single standard method for all cases. The importance given to physical function metrics highlights potential benefits from including rehabilitation as part of long-term management for heart failure.
Dr. Yamada stated: "Our findings reveal that physical function at discharge is a critically important determinant of survival, rivaling the importance of traditional cardiovascular risk factors. This study underscores the essential value of integrating comprehensive geriatric and functional assessments into the routine management and risk stratification of older patients with HF."
Researchers plan further testing before broader adoption but have begun developing a tool based on their Top-20 XGBoost model so clinicians can estimate individual patient risks more accurately.
