Japanese Team Builds XGBoost One-Year Mortality Model

Researchers at Juntendo University applied eXtreme Gradient Boosting to the nationwide J‑Proof HF registry (9,700 patients discharged Dec 2020–Mar 2022) to predict one‑year mortality, publishing findings Feb 3, 2026 in The Lancet Regional Health—Western Pacific. Their Top‑20 XGBoost model included seven physical‑function/non‑cardiac variables and improved risk reclassification by about 20% versus AHEAD and BIOSTAT compact, enabling more accurate identification of high‑risk elderly Japanese HF patients for targeted post‑discharge care.