11th International Conference on ICT for Sustainable Development

Authors - Ansh Soni, Aneri Shah, Krish Modi, Nishant Doshi
Abstract - This paper expresses an innovative method of data science architecture for enhanced diabetes prediction, embedding comprehensive mathematical framework, statistical feature engineering, and machine learning[I]. Utilizing the Pima Indian Diabetes dataset, we design and extract key features that drive prediction - Insulin-to-Glucose Ratio (IGR), Diabetes Risk Index (DRI), Metabolic Syndrome Score (MSS) and more—to analyze complex clinical dynamics. A detailed attempt of comparison study across logistic regression, decision trees, random forests, and deep neural networks enhanced by tuning model parameters [II] - indicates accuracy, precision.. By integrating mathematical and statistical methodologies, this methodology advances early diabetes detection, supporting the revolutionary impact of AI-driven analytics in custom healthcare.