A Predictive Model for Type 2 Diabetes Using Optimization Techniques and Deep Learning Models

Abstract

This research addresses the challenge of improving predictive accuracy in diabetes diagnosis and risk assessment by leveraging advanced deep learning techniques. The study primarily focuses on the detection of Type 2 diabetes and its complications such as diabetic retinopathy and coronary heart disease, employing the core methodologies to enhance predictive capabilities. The first approach utilizes deep learning models to analyze patient data, enabling the identification of individuals at risk with greater precision. By refining feature extraction processes and optimizing neural network architectures, this method significantly improves the classification of patients into low, medium, and high-risk categories, contributing to more targeted and effective diabetes management and also the percentage of damage of risk level. The second approach introduces a synergistic integration of deep learning with matrix factorization techniques, further enhancing the model s ability to capture complex patterns in patient data. This results in a more accurate forecast of diabetes onset and progression in the identification of risk levels, allowing for timely interventions and lowering the total impact of Type 2 diabetes on healthcare systems. The study leverages data from three comprehensive datasets: the National Health and Nutrition Examination Survey (NHANES), the UK Biobank, and the Framingham Heart Study. These datasets are employed to analyze diabetes and its associated risks, including diabetic retinopathy (DR) and coronary heart disease (CHD). The models, TreeFeatNet and HybridFeatureNet, are utilized to identify the presence of diabetes and the risk of complications such as DR and CHD. These models also predict the percentage of damage to the eye and heart, offering valuable insights into the severity of these complications in diabetic patients. Beyond diabetes diagnosis, the research expands into the analysis of diabetic complications, particularly diabetic retinopathy and coronary heart disease. The MESSIDOR, IDRiD...