A Simple Blood Test Could Reveal Hidden Radiation Exposure

Scientists at Lawrence Berkeley National Laboratory (Berkeley Lab) have created a machine learning tool that can analyze a standard blood count test to detect and estimate past exposure to radiation. This process, known as biodosimetry, uses biological changes to determine whether radiation exposure occurred and, if possible, estimate the amount of radiation absorbed.

Current methods for this are time-consuming and often inconclusive, especially for low-dose exposure which is more common. The researchers believe that a CBC-based model could serve as an accessible initial screening tool for radiation exposure. They were surprised to find that routine blood tests retain information about a person's prior radiation exposure across a wide range of doses and times after exposure.

The model, called Sparse CBC, has been made open source and is expected to be adaptable for human use. However, genetic background can significantly influence the radiation response signature present in CBC data, meaning future models should include a more diverse genetic population to improve generalizability and facilitate translation to human populations.