Machine-learning-based risk assessment of hypertension among older adults in Korea

South Korea is expected to become a super-aged society at the fastest rate among all Organization for Economic Co-operation and Development nations. Such a society can become a burden on primary healthcare, resulting in an increasing prevalence of chronic diseases, including hypertension (HYP). However, studies on the prevalence of HYP in older adults are lacking. This study aimed to construct an optimized performance machine learning (ML) model to assess HYP risk and identify associated risk factors among older adults in Korea.

This cross-sectional study included 3214 participants aged ≥65 years who were enrolled in the "Maeum Juchieui" program in Seoul, Korea. Depression, stress, anxiety, basal metabolic rate, oxygen saturation, heartrate, and average daily step count were measured using a digital device via a smartphone app. HYP, diabetes, hyperlipidemia, chronic obstructive pulmonary disease, body fat percentage, and body muscle percentage were also measured.

Of the five ML models, the Random Forest model exhibited the best performance. Hyperlipidemia, diabetes, basal metabolic rate, body muscle percentage, body fat percentage, average daily step count, stress, heartrate, and depression influenced HYP among older adults.

The findings indicate that health programs implemented by the Korean government for older adults that are directly or indirectly associated with HYP risk factors can have a positive effect on reducing HYP. Furthermore, the South Korean government must monitor policy-based programs for chronic diseases among older adults to ensure that such programs can be applied to those residing in vulnerable areas.