Hyperparameter Optimization Techniques for Enhanced Diabetes Prediction Using XGBOOST

Abstract

Data mining is crucial in healthcare since there is a mountain of data involved in disease diagnosis and analysis. Data analysis becomes exponentially more difficult under these circumstances, although they are not insurmountable. Health datasets are complicated and fraught with uncertainty; furthermore, they are arduous to manage and manipulate. One of the most significant healthcare issues impacting millions of people across the globe is diabetes. Diabetic early detection and prediction is crucial for initiating treatment in the early stages of the disease. Recent years have seen an exploration of machine learning in healthcare with the goal of assisting providers in making more accurate diagnoses and predictions about patient outcomes. In order to better anticipate cases of diabetes, this study investigates hyperparameter tuning with the Whale Optimisation Algorithm (WOA) in conjunction with the XGBoost machine learning method. The proposed approach utilizes 768 patient records from the Pima Indian Diabetes dataset in an effort to improve the efficiency and accuracy of disease prediction. Among the methodical processes included in the research are data preparation, hyperparameter identification, and WOA optimization. The optimized model shows encouraging accuracy and predictive performance outcomes when evaluated on a different dataset. Improving healthcare outcomes through the use of advanced machine learning and optimization techniques is the central focus of the research, as summarized in the abstract.