Stochastic expansion of radionuclide inhalation dosimetry for consequence management application: uncertainty and sensitivity analysis in the ICRP 130 human respiratory tract model.

Abstract

Releases from nuclear or radiological security events can result in significant internal radiation contamination through inhalation of particulate contaminants. The reference human respiratory tract model (HRTM), developed by the International Commission on Radiological Protection (ICRP) and described in the ICRP Publication 66 with subsequent updates in the ICRP Publication 130, is used to evaluate the deposition and clearance behaviour of inhaled radionuclides. Biokinetic models are used to quantify the retention and excretion of internally deposited particulates, supporting the calculation of inhalation dose coefficients (DCs). The HRTM developed by the ICRP utilises deterministic quantities outlined in the ICRP Publication 66 and 130. The overarching goal of this study was to determine the variability from deterministic biokinetic/dosimetry models to represent the stochastic breadth of radionuclide metabolism in an exposed occupational population from realistic source terms, yielding an expanded compendium of inhalation DCs. The analysis was carried out in three phases: (1) Implementation and extension of the biokinetic and DC modelling framework based on the ICRP Publication 130 HRTM and associated element specific systemic biokinetic models; (2) Investigation of uncertain parameters in the HRTM; and (3) Stochastic analysis using Latin hypercube sampling, incorporating non-parametric (Kolmogorov-Smirnov statistics) test and quantile-quantile plot assessment to support parametric distribution selection, for characterisation of the committed effective DCs distributions. To determine the most impactful parameters among the uncertain parameters, a random forest regression model was employed for feature importance, coupled with SHapley Additive exPlanations for comprehensive machine learning interpretation of the features. This study presents a unique stochastic framework for modelling inhaled particulate metabolism, enhancing capabilities in radiation consequence management, medical countermeasure development, and radiation dose reconstruction for epidemiological investigations.