An Efficient Approach to Detect Diabetes using XGBoost Classifier
Kumar Laxmikant, R. Bhuvaneswari, Balasubramaniam Natarajan
Abstract
Diabetes is a chronic and metabolic health condition that has overwhelmed a myriad of people worldwide. It is characterized by the elevated blood glucose level, which may lead to serious damage to organs like heart, eyes and kidney. Besides that, having an unhealthy and sedentary lifestyle have made it a lifestyle disease. With the advancement in the field of Artificial Intelligence (AI) and Machine Learning (ML) techniques, early detection of diabetes is being automated. Various Machine Learning techniques were analysed in this work to conduct a thorough relationship study on multiple diabetes-related factors such as Blood/Systolic Pressure, Thickness of the skin, Pregnancy, Insulin, BMI, Age, and so on. After a comparative study, the Extreme Gradient Boosting (XGBoost) Algorithm has been proven to be the best model due to its high performance. As the proposed model has good generalizability, its accuracy is 94.8%, which is a 7.2% improvement on existing techniques. The classification performance has been evaluated by assessing precision, recall, F1-score, and the receiver operating characteristic area under the curve (AUC) which also shows an impressive results.