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

(1) Background: Various machine learning techniques were used to predict hypertension in Korean adults aged 20 and above, using a range of body composition indicators. Muscle and fat components of body composition are closely related to hypertension. The aim was to identify which body composition indicators are significant predictors of hypertension for each gender; (2) Methods: A model was developed to classify hypertension using six different machine learning techniques, utilizing age, BMI, and body composition indicators such as body fat mass, lean mass, and body water of 2906 Korean men and women; (3) Results: The elastic-net technique demonstrated the highest classification accuracy. In the hypertension prediction model, the most important variables for men were age, skeletal muscle mass (SMM), and body fat mass (BFM), in that order. For women, the significant variables were age and BFM. However, there was no difference between soft lean mass and SMM; (4) Conclusions: Hypertension affects not only BFM but also SMM in men, whereas in women, BFM has a stronger effect than SMM.