1인 가구와 다인 가구의 연령대별 위장질환 발생률 예측 모델

Predictive Model for Age-specific Incidence Rates of Gastrointestinal Diseases in Single-Person and Multi-Person Households

This study employed machine learning and deep learning regression models to predict the incidence of gastrointestinal diseases. In particular, attention was given to the increasing number of single-person households, emphasizing that irregular eating habits and dining out may contribute to gastrointestinal health problems. The incidence of gastrointestinal diseases was predicted and compared between single-person and multi-person households by age group, with categorical variables such as age, gender, disease, and household type converted into numerical data through one-hot encoding during the data preprocessing phase. The regression models used for prediction included linear regression, random forest, support vector machines (SVM), gradient boosting, XGBoost, and multilayer perceptron (MLP). After constructing the models, their performance was evaluated to identify the most suitable one. The evaluation results indicated that the XGBoost regression model provided the best performance. Using this model, predictions of gastrointestinal diseases by age group for both single-person and multi-person households were made and visualized. The findings highlight the need for dietary improvement programs and policy development for single-person households and are expected to contribute to the formulation of future public health policies.