Age-Related Differences in Risk Factors for Long-Term Care Certification in Japan: A Decision Tree Analysis Spanning 3 Years

Ensuring healthy aging and extending health span is critical, particularly in aging societies. This study analyzed patterns of risk for long-term care (LTC) certification within 3 years in rapidly aging areas of Japan using the Kihon Checklist (KCL) and machine learning models. Data from adults aged 65 years or older in Iiyama City (aging rate 37.0%) were analyzed using Exhaustive Chi-squared Automatic Interaction Detector decision trees. The dependent variable was LTC certification, and independent variables included age, sex, and six KCL domains: physical strength, nutrition, oral function, isolation, memory, and mood. Three risk patterns demonstrated consistent results across training and evaluation datasets. Age was the strongest determinant of LTC certification. Among those aged 80 years or older, low cognitive function and depression were key risk factors, while younger groups showed stronger associations with physical weakness. Results suggest age-specific segmentation of populations is crucial for designing effective interventions to prevent LTC certification.