Estimation of low-level net counting rates and exposures

Bayesian statistical techniques have the advantage that information other than data can be incorporated into the analysis. In health physics, these techniques are most useful when the available data are limited. Two examples using these techniques will be presented. In the analysis of low-level counting data, the classical techniques sometimes yield negative net counting rates even when it is clear, from other extraneous information that the net rate should be positive. Using such information in Bayesian estimation results in positive net rates and the credible intervals do not include negative values. Software using this technique was prepared and the method is compared with the classical method using urinalysis data. It is the first time that this technique was used to estimate the net rate for uranium urinalysis data. Another example for the Bayes approach is the derivation of the calibrative density for a linear regression model. A computer program for this techniques was written to infer an unknown dose from the dosimeter calibration data and the dosimeter readout. Examples for the calibrative density based on TLD data are shown. It is the first application of the calibrative density to a problem in external radiation dosimetry.