Application of principal component regression at the U.S. Transuranium and Uranium Registries

 On Monday September 19th at 4pm in Olin 201 Martin Šefl will talk about the method of principle component analysis (PCA) in the context of regression on data from the United States Transuranium and Uranium Registries (USTUR).

The United States Transuranium and Uranium Registries (USTUR) studies actinide (Pu, Am, and U) biokinetics and tissue dosimetry by following up former nuclear workers with documented internal deposition of these radioactive elements. The USTUR research relies heavily upon postmortem autopsy findings and radiochemical analyses of donated tissues. Plutonium is a radioactive element first produced and isolated in the 1940s. The skeleton is a major plutonium retention site in the human body. The estimation of the total plutonium activity in the skeleton is a challenging problem. For most tissue donors at the USTUR, a limited number of bone samples is available. Typically, the total skeleton activity is calculated using plutonium activity concentration (C_skel) and skeleton weight. If limited number of bone samples was analyzed, C_skel could be estimated using multiple linear regression (MLR) of data from whole-body donors, where C_skel were estimated based on the analysis of the half of the skeleton. The caveat of MLR is that individual bone sample concentrations are correlated. Multicollinearity can be addressed by principal component regression (PCR). PCR consist of two steps, PCA and regression. PCA enables to reduce the number of variables for regression while retaining information. A brief introduction into PCR and its application of PCR for estimates of plutonium concentration in the skeleton will be discussed.

Dr. 

BIO: Martin Šefl is a Post-doctoral Research Associate at the United States Transuranium and Uranium Registries, College of Pharmacy and Pharmaceutical Sciences, Washington State University. His research is focused on biokinetic modeling and radiation dose calculations of plutonium internally deposited in the human body. Dr. Šefl holds a PhD degree in Radiological Physics from the Czech Technical University in Prague. His doctoral research was focused on the cellular radiation dosimetry calculations using Geant4 Monte Carlo code. He worked as a medical physicist in the department of radiation oncology in a hospital in Prague.