Diabetes detection has become a critical area of research due to the rapid global increase in diabetic cases. This paper proposes a hybrid model integrating object-oriented software design with machine learning (ML) and deep learning (DL) techniques to predict diabetes with improved accuracy and scalability. A structured software design using UML diagrams supports modular development, while advanced ML and DL algorithms enhance prediction performance. We used datasets with a focus on non-invasive features and applied preprocessing, feature selection, and algorithm tuning. The CNN and SVM models were evaluated, achieving 94% and 86% accuracy, respectively. The system shows promise for real-world healthcare applications, especially in developing costeffective and efficient mobile-based diagnostic tools.
Volume 17 | Issue 4
Pages: 1-4