Diabetes Prediction using Diabetic Biomarkers: A Literature Survey

Diabetes is a growing global health concern requiring early detection and continuous monitoring. In this work, we address key research gaps identified from the review of existing studies, including the lack of real-time biomarker integration, transparency in prediction, and personalized recommendations. We propose a multi-source diabetes prediction system that processes data from clinical reports, wearable devices, manual inputs, and electronic health records. The system utilizes NLP techniques for biomarker extraction and employs a machine learning model to classify patients as diabetic, pre-diabetic, or non-diabetic. An Explainable AI (XAI) layer is integrated to interpret model decisions and highlight key biomarkers contributing to the prediction. Additionally, the system provides personalized health recommendations based on the analysis. This approach aims to deliver an accurate, real-time, and patient-centric solution for effective diabetes prediction and management.