AI-driven triage classification at the 1 Gy threshold using dietary supplements and portable OSL dosimetry

In radiological emergency scenarios, the rapid distinction between individuals exposed below or above the 1 Gy triage threshold is essential for effective medical sorting and optimized resource allocation. This study proposes a binary classification framework at the 1 Gy threshold that combines Optically Stimulated Luminescence (OSL) measurements of commercial magnesium-based dietary supplements with supervised machine learning (ML) algorithms, including Logistic Regression, Support Vector Machines, Decision Trees, Random Forest, and XGBoost. A dataset of 355 samples exposed to different radiation doses was analyzed, and model performance was evaluated using cross-validation and multiple statistical metrics. The resulting framework was implemented into a lightweight, browser-based application to provide real-time predictions and support decision-making in field operations. The findings demonstrate that integrating physical dosimetry with ML enables rapid and scalable classification relative to the 1Gy threshold and offers a practical tool to enhance public health response during radiological incidents.