A novel ensemble machine learning framework for early stage diabetes mellitus prediction

Abstract

Diabetes Mellitus is a harmful condition characterized by elevated blood sugar levels resulting from insufficient insulin production in the body. This disorder gives rise to various health complications affecting the kidneys, heart, nerves, and eyes. If left unidentified and untreated, it can even prove fatal. Therefore, early detection and prediction of chronic diseases are imperative, and advancements in technology have led to a shift towards personalized healthcare. Machine learning offers a viable solution for predicting the likelihood of developing the disease with satisfactory accuracy.The objective of this research is to create a machine learning-based model for accurately predicting early-stage diabetes mellitus. The study employs PCA for dimensionality reduction and the ensemble bagging decision tree classification, thereby, achieving a remarkable accuracy level concerning the disease. The proposed model is evaluated using a publicly available dataset of 520 instances. The dataset utilized in this research includes information such as Polyuria, Polydipsia, sudden_weight_loss and weakness etc. The model achieves an impressive accuracy of 93.26%, precision of 96.9%, recall of 94.1%, and F-Score of 95.4%. Comparative analysis against other existing techniques demonstrates its superior performance in predicting such chronic conditions.