Predict Diabetes Complications Using the Improved Grey Wolf Optimizer Algorithm and SVM

Diabetes complications are among the leading causes of mortality worldwide, making their accurate prediction crucial. The complexity and similarity of symptoms often make diagnosis challenging. This paper proposes an enhanced Grey Wolf Optimization (GWO) algorithm, integrated with an adaptive median filter, to optimize the parameters of Support Vector Machines (SVM) for improved prediction accuracy. The proposed method dynamically filters out outliers during optimization, improving accuracy and reducing errors. By combining the GWO’s global search capability with the filtering precision of the adaptive median filter, the algorithm avoids local optima and achieves superior classification performance. Experimental results demonstrate that the proposed approach outperforms conventional methods, including Decision Trees (DT), Bayesian Classifiers (BC), and Neural Networks (NN), in terms of accuracy and error reduction.

Keywords:

Diabetes , Support Vector Machine , Grey Wolf Optimization , Adaptive Median Filter , Classification