Prediction and Diagnosis of Diabetes Mellitus Classes in Humans using Support Vector Machine

Diabetes mellitus (DM) has become a chronic condition marked by intentional production of aberrant glucose levels in the bloodstream resulting from food culture in recent years. The main causes of diabetes development are inadequate insulin production and cell resistance to its actions. < Thus, it is imperative to use machine learning techniques to automatically identify diabetes mellitus presence. Thanks to the development of machine learning, recent works have made it feasible to generate exact and accurate predictions of DM. Still, the presence of redundant data causes most of these forecasts to have low prediction accuracy. In this work, we predict the classes of DM by means of an integrated mechanism of feature selection and classification technique employing support vector machine (SVM). This mechanism also comprises a method for feature selection. SVM helped to significantly cut the computing time needed. This was so because the suitable selection of instances made possible by the effective feature selection. Running the simulation against the Pima Indians Dataset, the outcomes show that the proposed SVM has a better rate of accuracy than other techniques now in use.