11 April 2025 in International Research Journal of Multidisciplinary Technovation
The rising prevalence of diabetes, driven by dietary changes and reduced physical activity, is a leading cause of mortality worldwide. Early diagnosis is critical to managing this chronic condition. This study proposes an AI-based intelligent system for early diabetes detection using Edge-Based Neural Random Backpropagation (EB-NRBP). The EB-NRBP model leverages feature selection via the Least Absolute Shrinkage and Selection Operator (LASSO) to enhance the regularization of the classification process. This approach optimizes the classifier’s cost function and accelerates its development through Random Forward Gradients in conjunction with Edge-Based neural networks. The model's performance was compared with conventional methods, demonstrating a significant improvement in classification accuracy. The EB-NRBP model achieved a high success rate of 98%, outperforming traditional techniques in terms of efficiency and precision. This AI-based system presents a promising solution for the early diagnosis of diabetes, offering higher accuracy and faster detection compared to existing methods. It holds potential for integration into healthcare applications, enhancing early intervention and improving patient outcomes.