“…In establishing an approach to use in addressing health risks for low radiation doses to humans, researchers need to take into consideration that unlike animal-studies data, epidemiologic-studies data for humans are very noisy (wider stochastic error distributions) 5 so that any risk estimates generated using such data likely involve large errors. 5 , 6 Animal-studies data with smaller errors (for radiation doses and biological effects) than epidemiologic-studies data could be used in improving modeling methods employed in epidemiologic studies and in testing reliability of cancer risk predictions for low radiation doses.…”