Machine Learning for Optimizing Diabetes Prediction: A Comparative Analysis

This article presents a comprehensive comparative analysis of machine learning algorithms for optimizing diabetes prediction. With a focus on detecting chronic diseases, particularly diabetes, we explore the performance of various machine learning models using a diabetes dataset collected from a hospital in Frankfurt, Germany. Our study spans the years 2020 to 2023, encompassing the latest advancements in the field. We employ well-known algorithms such as Logistic Regression, Decision Tree, Random Forest, and Multi-layer Perceptron, evaluating their accuracy and F1 score as primary metrics. Additionally, we incorporate data cleaning techniques, KFold cross-validation, and hyperparameter tuning to enhance model performance. Our findings shed light on the effectiveness of these algorithms and the significance of data cleaning in predictive modelling. We also investigate the optimal number of trees in a Random Forest for diabetes detection. Through this rigorous analysis, we aim to contribute to the advancement of both machine learning and healthcare by improving disease detection methods.