Machine Learning Approach for Predicting Hypertension Based on Body Composition in South Korean Adults

Area of Science:

• Cardiology
• Biomedical Engineering
• Data Science

Background:

• Hypertension is a significant health concern in adults.
• Body composition, particularly muscle and fat, is linked to hypertension risk.
• Predictive models can aid in early hypertension detection.

Purpose of the Study:

• To identify significant body composition predictors of hypertension in Korean adults.
• To compare predictive accuracy across different machine learning techniques.
• To determine gender-specific predictors for hypertension.

Main Methods:

• Utilized six machine learning techniques to classify hypertension.
• Included age, BMI, body fat mass, lean mass, and body water data from 2906 Korean adults.
• Developed a predictive model for hypertension classification.

Main Results:

• Elastic-net model achieved the highest classification accuracy.
• For men, age, skeletal muscle mass (SMM), and body fat mass (BFM) were most predictive.
• For women, age and BFM were significant predictors; SMM and soft lean mass showed no difference.

Conclusions:

• Hypertension is associated with both body fat mass and skeletal muscle mass in men.
• Body fat mass has a greater impact on hypertension in women compared to skeletal muscle mass.
• Machine learning effectively identifies key body composition indicators for hypertension prediction.