Biomarker Dataset for Prediabetes Classification Using Interpretable Machine Learning

This dataset contains clinical and biomarker measurements from 604 adult participants who attended the DiabHealth rural diabetes screening clinic between 2002 and 2015. Variables include demographics (Gender, Age), cardiovascular history (CVD-Revised, HT-Status), glycemic markers (ScreenGlucose, HbA1c), lipid profile (Triglyceride, TC, HDL, LDL), inflammatory markers (CRP, IL-6, IL-1Beta, IL-10, MCP-1, IGF-1), oxidative stress markers (8-OHdG, GSH, GSSG, GSH/GSSG), and mitochondrial-related peptides (Humanin, MOTS-c, p66Shc). Prediabetes and control status in the associated manuscript were derived from ScreenGlucose according to ADA fasting glucose criteria (5.6–6.9 mmol/L for prediabetes; <5.6 mmol/L for controls). All data are de-identified and were used to develop and validate interpretable machine learning models for prediabetes classification.