A simple algorithm to predict disability in community-dwelling older Japanese adults

Area of Science:

• Gerontology and Public Health
• Computational Health Informatics

Background:

• Global aging population necessitates proactive disability prevention strategies for older adults.
• Predicting disability risk is essential for timely and effective interventions.
• Mobile health applications offer a promising avenue for remote assessment and personalized care.

Purpose of the Study:

• To develop a predictive model for disability risk in older adults.
• To utilize decision tree analysis for identifying key risk factors.
• To inform the development of mobile applications for disability risk assessment and intervention prioritization.

Main Methods:

• Analysis of a cohort study involving 12,000 older adults without pre-existing disabilities.
• Application of the Chi-square automatic interaction detection (CHAID) algorithm for decision tree analysis.
• Identification of partitioning variables and end nodes to stratify disability risk.

Main Results:

• The CHAID model achieved 73.4% classification accuracy and an area under the curve of 0.76.
• Identified 24 end nodes and 16 partitioning variables from 34 questionnaire items.
• Disability risk probabilities ranged from 0.0% to 96.7%, varying by age group and factors like mental health, lifestyle, and geriatric syndromes.

Conclusions:

• Disability risk factors and their influence vary significantly across different age groups of older adults.
• Mental health is a key factor for those in their 80s and older.
• Lifestyles and geriatric syndromes are critical for predicting disability in those in their 70s, informing targeted interventions.