A comparative study of the diabetes progression prediction techniques

Diabetes is a global health crisis affecting millions all over the world. The early and accurate prediction of diabetes plays a crucial role in preventing severe complications and improving patient outcomes. Regression-based machine learning techniques have emerged as powerful tools for diabetes prediction, enabling the modeling of complex relationships between risk factors and disease progression. As these algorithms continue to advance, incorporating better feature selection, regularization, and ensemble methods, their predictive performance has significantly improved, enabling more precise and personalized healthcare interventions. Despite these advancements, the rapid proliferation of regression techniques, datasets, and evaluation metrics has made it increasingly difficult to compare different approaches fairly. To address this challenge, this paper presents a comprehensive survey of current trends in diabetes prediction, with a focused analysis of regression models. The primary objectives of this paper are as follows: 1) Investigate the performance and evolution of regression-based machine learning techniques, assessing how algorithmic enhancements improve prediction accuracy and reliability. 2) Establish a standardized reference framework for evaluating and benchmarking diabetes prediction models, ensuring fair and reproducible comparisons. 3) Evaluate the effectiveness of leading regression algorithms in this domain, identifying which methods best balance interpretability and predictive power for healthcare applications. 4) Outline future research directions, emphasizing the need for more robust, explainable, and patient-centric prediction models that can seamlessly integrate into clinical practice.