PREDICTIVE MODELING FOR DYSLIPIDEMIA RISK USING SELF-MEASURABLE PHYSICAL FACTORS: A MACHINE LEARNING APPROACH

초록

Dyslipidemia, characterized by abnormal lipid levels in the blood, is a major risk factor for cardiovascular diseases. Despite its high prevalence, public awareness of dyslipidemia remains low, and many individuals are unaware of their risk until they develop serious health complications. This study proposes a predictive model for dyslipidemia based on self-measurable physical factors, aiming to enable early detection and prevention. Using machine learning algorithms, including Decision Tree, Random Forest, XGBoost, and ensemble methods (Voting and Bagging), the model is developed with data from the National Health Insurance Service’s Health Examination dataset. The dataset includes variables such as age, BMI, waist circumference, smoking and drinking habits, and cholesterol levels. The model’s performance is evaluated using metrics such as precision, recall, F1-score, accuracy, and Area Under the Curve (AUC). Among the models tested, the Voting Ensemble model demonstrated the best performance, with a precision of 0.862, recall of 0.852, F1-score of 0.851, AUC of 0.927, and accuracy of 0.85. These results suggest that the proposed model could be a valuable tool for individuals to assess their risk of dyslipidemia, thereby enhancing personal health management and contributing to the prevention of cardiovascular diseases. Future work will focus on validating the model with diverse datasets and exploring its applicability in real-world settings. © 2025, ICIC International. All rights reserved.

키워드