Author(s) :

Ava Ng, Junior.

Volume/Issue :

Abstract :

The present study addresses the escalating public health concern of diabetes, which affects approximately 34.2 million Americans. Recognizing the multifaceted etiology of diabetes, encompassing genetic, environmental, and lifestyle factors, the research aims to identify influential determinants of this chronic condition. A comprehensive dataset of 100,000 entries from Kaggle is utilized, thereby embracing the advantages of observational data analysis. The study employs a hybrid methodological framework, integrating machine learning and traditional statistical techniques. Specifically, the Random Forests for feature importance analysis and binary logistic regression are leveraged to understand the relationships between variables and diabetes risk. This dual approach allows us to harness the predictive power and variable ranking capability of machine learning while maintaining the interpretability and statistical rigidity of logistic regression. The analysis encompasses a broad range of factors, including biological aspects like BMI and blood sugar levels, and other socio-economic determinants. By combining diverse methodologies, this study intends to provide a more detailed understanding of diabetes risk factors, facilitating the development of targeted prevention strategies and informing policy decisions. The findings hold the potential to significantly impact health education, innovative healthcare solutions, and policy development, addressing a critical need in diabetes management and prevention.