A REVIEW OF MACHINE LEARNING TECHNIQUES FOR DIABETES RISK PREDICTION

Abstract

Diabetes mellitus is one of the most common chronic diseases affecting millions of individuals worldwide and has become a major challenge for healthcare systems. The increasing prevalence of diabetes has created a strong demand for early prediction and preventive healthcare solutions. Traditional diagnostic procedures mainly rely on laboratory investigations and clinical expertise, which may not always support early identification of high-risk individuals. In recent years, machine learning techniques have emerged as promising tools for predictive healthcare due to their capability to analyze medical datasets and identify hidden disease patterns. Machine learning algorithms can assist healthcare professionals in predicting diabetes risk using clinical and physiological parameters such as glucose level, insulin concentration, body mass index, age, blood pressure, and hereditary factors. Several classification techniques, including Logistic Regression, Decision Tree, K-Nearest Neighbors, Random Forest, and deep learning models, have been applied for diabetes prediction. However, challenges related to prediction accuracy, feature relevance, computational complexity, overfitting, and interpretability continue to affect model performance. This review paper presents a detailed survey of machine learning techniques used for diabetes risk prediction. Existing studies are critically analyzed based on methodologies, datasets, predictive performance, strengths, and limitations. A comparative evaluation of commonly used classification algorithms is also provided to identify research trends and current limitations. The findings indicate that although intelligent techniques significantly improve prediction capability, there is still a need for computationally efficient, interpretable, and feature-optimized models suitable for practical healthcare applications. The study further highlights important research gaps and future directions for developing improved diabetes prediction systems.