Enhanced SVM Classification for Diabetes Prediction: A Comparative Analysis Using the Kaggles Diabetes Dataset

Abstract

Diabetes mellitus is a significant global health concern that impacts a large number of individuals globally and imposes a substantial financial burden on healthcare systems. The aim of this study is to use machine learning methods, namely Support Vector Machines (SVM), to develop a prediction model for assessing the risk of diabetes using the Kaggle diabetes dataset. We used a comprehensive dataset sourced from Kaggle, which encompasses several health metrics such as age, body mass index (BMI), glucose levels, and other relevant factors. In order to identify patterns that indicate a potential risk of diabetes, our approach included doing data pre-processing, selecting relevant features, and implementing a Support Vector Machine (SVM) classifier. In order to assure the strength and reliability of the SVM model, it was trained and validated using standard cross-validation techniques. We evaluated its performance by using F1-score, accuracy, precision, and recall criteria. Based on our findings, the SVM approach shows promise in predicting the risk of diabetes, with an accuracy of 83.12% on the test set. Despite the encouraging findings, we acknowledge the need for future improvement of the model and the limits of our work. Subsequent investigations might use deep learning techniques or ensemble approaches to enhance the accuracy of predictions. This work contributes to the growing body of research on the use of machine learning in healthcare and has the potential to influence strategies for early identification and prevention of diabetes. Prior to considering actual implementation, more clinical validation is necessary. This introduction, written in APA format, specifically examines studies that used the Kaggle diabetes dataset or similar datasets for the purpose of predicting diabetes. It includes a minimum of 12 references.