Enhancing Diabetes Risk Prediction with Hybrid Machine Learning Models

This paper explores the integration of causal inference with machine learning (ML) to enhance early diagnosis and effective management of diabetes. By leveraging advanced techniques such as data preprocessing, causal analysis, evaluation of variable importance, feature engineering, and hyperparameter optimization, we develop a predictive model using a Stacking ensemble that combines multiple base models. Initial results demonstrate significant improvements in model performance, suggesting that this integrated approach offers a promising direction for diabetes management.