Diabetes Prediction Using Machine Learning: Methods, Challenges, and Insights from a Systematic Literature Review (SLR)

This study systematically reviews and critically evaluates the current state of machine learning models for diabetes prediction, addressing key methodologies, challenges, and insights. Despite the growing body of research, a comprehensive systematic literature review (SLR) on this topic has been lacking. By consulting five major scientific databases (IEEE, ScienceDirect, Springer, Scopus, and ACM), this paper offers an in-depth analysis of existing studies' strengths, limitations, and research gaps. Key challenges discussed include data collection and preprocessing, handling missing values, feature importance assessment, standardization, and addressing class imbalance in datasets. Additionally, the review identifies underexplored areas and highlights opportunities for future research, such as developing standardized preprocessing frameworks and exploring advanced hybrid models. This SLR aims to guide researchers by summarizing existing evidence, resolving conflicts in the literature, and providing actionable directions for advancing diabetes prediction through machine learning.