Effectiveness of Machine Learning Models and Performance Enhancement with Threshold Tuning Method Adopted in Diabetes Prediction

Using machine learning algorithms to predict diabetes has been the focus of extensive research by a few researchers. The objective of this paper is to provide PWDs, clinicians, family members, and carers with a better understanding of the possible uses of current AI advancements in the treatment of PWDs. Millions of PWD could benefit from the enhanced blood glucose management, lower frequency of hypoglycemic episodes, and decreased risk of diabetes-related comorbidities that could be facilitated by AI applications. AI applications enhance the lives of people with disabilities (PWDs), by enhancing accuracy, efficiency, usability, and satisfaction. In this study, the performance metrics of the top three machine learning models currently available are compared. Four out of the five parameters analyzed, the KNN model fared the best. KNN is superior in terms of accuracy, area under the curve (AUC), precision, and f1 score. Logistic Regression achieves the optimal Recall/Sensitivity of 93%.