Radiation Biomarker Screening and Dose Reconstruction based on Machine Learning

Abstract: To address the urgent need for rapid and wide-range radiation biodosimetry in nuclear emergency medical response, this study established a powerful machine learning framework for biomarker screening and dose reconstruction. A multi-stage screening strategy integrating differential expression analysis, Spearman correlation filtering and Boruta algorithm was employed to identify 172 highly robust biomarkers from an initial pool of 25,654 genes. By utilizing a stacked ensemble machine learning approach, high-accuracy dose reconstruction was achieved across a broad range of 0-12 Gy, with an R² of 0.991 and an RMSE of 0.414 Gy, significantly outperforming conventional regression analysis and individual machine learning models. Further refinement reduced the gene panel to just 11 key markers while preserving reconstruction accuracy comparable to the full 172-gene model. Experimental validation via qRT-PCR confirmed the reliability of these biomarkers, demonstrating the framework’s potential for translation into a field-deployable diagnostic platform for radiation exposure assessment.