Immuno-Redox Signatures in Heroin–Methamphetamine Co-Use - Supplementary Files

This dataset contains individual-level immune-redox measurements and machine learning classification outputs used to distinguish heroin-only users from heroin-methamphetamine co-users in a Sri Lankan pilot cohort. Variables include normalized values of IL-6 and IL-4, the Th1:Th2 cytokine ratio (IL-6:IL-4), antioxidant capacity (ABTS), nitric oxide metabolites (NOx), and the oxidative stress index (OSI). The dataset also includes clinically defined group labels, artificial neural network (ANN) model-assigned group classifications, and predicted class probabilities for heroin and heroin–methamphetamine categories generated under leave-one-out cross-validation. These data enable evaluation of multivariate group separability and concordance between clinical labels and model-based classifications.