Machine Learning Analysis of Health and Lifestyle Factors in Understanding Diabetes

Abstract

Diabetes poses a significant global health challenge, with approximately 537 million adults living with the condition in 2021, a number expected to rise to 783 million by 2045. To enhance predictive accuracy and gain deeper insights into the factors contributing to diabetes, this study employed machine learning algorithms to predict diabetes risk factors using a dataset encompassing health and lifestyle variables. Six supervised machine learning algorithms, including Gradient Boosting, Logistic Regression, and Random Forest, among others, were assessed for their effectiveness in classifying diabetes status into two categories: diabetes and no diabetes. The study found that Gradient Boosting achieved the highest overall accuracy at 85%, demonstrating the best recall for diabetic cases at 57%. Meanwhile, Logistic Regression excelled in precision for non-diabetic cases at 94%. Key risk factors identified include general health status, blood pressure, body mass index, cholesterol levels, and age. Notably, the study uncovered that higher income and education levels were associated with increased diabetes risk, contradicting some existing literature and indicating the potential impact of lifestyle factors.