Hybrid Machine Learning Models for Type-3 Diabetes: Advancing Classification and Forecasting Environments

Diabetes Mellitus (DM) is a prevalent metabolic disorder that necessitates accurate diagnosis and forecasting for effective management. This paper examines predictive modeling techniques for diabetes classification and Type-3 Diabetes (T3DM) prediction using the NC State University Diabetes Dataset and the Ohio Longitudinal 2023 Dataset. By employing various machine learning approaches, data preprocessing, and exploratory data analysis, we achieve an 81.1% accuracy rate in diabetes classification through a hybrid method. This approach combines the predictive strengths of gradient boosting and logistic regression classifiers. Additionally, we develop a T3DM prediction model that effectively identifies individuals at risk, showcasing its strong predictive capabilities. The implementation of the hybrid algorithm highlights its superior performance compared to standalone models. Our results emphasize the importance of predictive modeling in improving diabetes management, aiding healthcare decision-making, and enhancing patient outcomes. Future research could explore incorporating additional data sources and advanced techniques to further improve the clinical relevance and predictive accuracy of diabetes care.