Bayesian prior probability distributions for internal dosimetry

Area of Science:

• Internal Dosimetry
• Bayesian Statistics
• Radiological Health

Background:

• Choosing appropriate prior distributions is crucial for Bayesian interpretation of internal dosimetry measurements.
• Historical bioassay data from Los Alamos provides a basis for evaluating prior distribution models.

Purpose of the Study:

• To theoretically analyze and empirically evaluate prior distribution models for Bayesian internal dosimetry.
• To propose and implement improved prior distribution models for tritium and plutonium bioassay data.

Main Methods:

• Theoretical analysis of prior distribution selection.
• Examination of historical tritium and plutonium urine bioassay data.
• Development and application of log-normal and alpha distributions as prior models.

Main Results:

• Two prior probability distribution models, log-normal and alpha (a gamma distribution variant), were proposed.
• These models were incorporated into version 3 of the Los Alamos Bayesian internal dosimetry code.
• Plutonium internal dosimetry now utilizes self-consistently determined prior parameters from population averages.

Conclusions:

• The proposed log-normal and alpha distributions offer viable options for prior distributions in Bayesian internal dosimetry.
• The integration into Los Alamos's code facilitates practical application and improved accuracy in dose assessments.
• Self-consistent parameter determination enhances the reliability of plutonium internal dosimetry at Los Alamos.