Personalized Diabetes Prediction Method Using Machine Learning Based on Expanded Health Data

This study proposes a personalized diabetes prediction model by employing the AdaBoost classifier, a machine learningbased artificial intelligence (AI) algorithm, to integrate diverse health indicators. Using data from the 2022 Korea National Health and Nutrition Examination Survey (KNHANES), the model incorporates extended variables—including age, body mass index (BMI), urine sugar, hepatitis A antibody, and marital status—in addition to traditional physiological diagnostic indicators such as fasting blood glucose and glycated hemoglobin (HbA1c). The proposed model achieved 78% performance in accuracy, precision, recall, and F1-score under 5-fold cross-validation, which is meaningful as it was obtained through multivariate analysis with extended health attributes. Compared to conventional models centered on traditional indicators, the proposed approach demonstrated improved balance and stability in predictions, thereby contributing to the early identification of high-risk groups and the establishment of prevention-focused healthcare strategies.