Deep Learning Algorithms for Diabetes Detection: A Comprehensive Exploration and Evaluation

Abstract

Early diagnosis of diabetes can lead to early interventions, lifestyle modifications, and personalized treatment plans that can positively impact patient health outcomes and reduce the burden on healthcare systems. Early detection reduces the risk to the patient’s health. Diabetes is spreading rapidly all over the world, and the majority of people over the age of 45 are victims. Therefore, it is important to detect this serious disease as soon as possible. Traditional diabetes screening methods often involve regular blood tests and clinical evaluations, which may not always detect diabetes in its early stages or identify individuals at high risk of developing the condition. A deep learning model is useful to detect this disease, which also reduces the cost of medical care. In this paper, we used different models, including LR, SVM, KNN, and NB, to analyze diabetes and then show their comparative results. Experiments are conducted on two datasets: the Pima Indian diabetes dataset and the Diabetes Health Indicator, both of which are available on Kaggle.