Statistical investigation of the random variations in PIXE hair analysis

Many studies have been performed to determine the associations between the amounts of minerals present in hair and the health status, toward developing new approaches for disease prevention and prediction. Unfortunately, because of the highly variable results, which can be caused by many factors, there is no definitive consensus on the validity of these values in disease management. A previous proton-induced X-ray emission (PIXE) study of mothers and infants examined the association between hair minerals and the onset of atopic dermatitis in infants. For many of the minerals identified and analyzed, highly variable intra-individual variations proved problematic, and may have obscured any connections between disease states and the mineral levels. We feel that better statistical modeling of these intra-individual variations might lead to a better understanding of the associations between the hair mineral levels and the patient health status. This report examines the statistical nature of the intra-individual variations due to locations of hair strands and those due to the physical errors of the PIXE experiments. The results for Cu, Fe, Sr and Se are presented.