A multidimensional statistical analysis of data obtained from breath gas measurements with Proton Transfer Reaction-Mass Spectrometry (PTR-MS) is proposed, based on a chemical-diffusion equilibrium approach. The proposed methodology is developed and demonstrated on the problem of detecting exposure of human beings to ionizing radiation. It could be applied to a general family of non-invasive, high-throughput, breath gas based detection strategies, like for instance a breath gas test for early diagnosis of lung cancer.