Citizen science approach to mapping background radiation for environmental epidemiology

BACKGROUND AND AIM Volunteered geographic information from citizen scientists can provide alternatives to traditional exposure assessment for environmental epidemiological studies; whereby, facilitating access to data, providing continuous monitoring, improving exposure awareness for participants and the community, among others. Safecast is a participatory, open-source, citizen science centered radiation surveying project where international volunteers collect and openly share data on environmental radiation collected through mobile sensors. However, information is lacking on the potential use of the collected measurements for epidemiological research, especially regarding their validity for exposure assessment. We aimed to develop and validate maps of gamma radiation (GR) for Denmark and Switzerland, as part of a larger European-wide consortium, using measurements collected by the Safecast community. METHODS Spatial Bayesian mixed-effects models were fit to GR measurements from Safecast, using predictors of soil class, geology, land cover, and elevation from maps. Pointwise posterior means from these models were used to predict GR for the territory of each country at a resolution of 1x1km2. Internal validity was assessed using cross-validation (CV) with random assignment of either the measurements (random-CV) or area blocks defined by square grids (different cell sizes) to folds (spatial-CV). We also validated maps for Switzerland against personal exposure measurements (dosimeters worn over five consecutive days) in 149 children using linear regression and calculated the coefficient of determination(R2) and mean squared error. RESULTS Spatial variations in GR exposure was observed for each country. Predictors of GR from Safecast were associated with ambient dose rates, where models fit the data well, for example with a R2 of 0.5 in random-CV, with 0.77 model fit in Denmark. CONCLUSIONS Results highlight the potential for using open citizen science data for exposure assessments in radiation research. The association between GR and childhood cancers will be assessed. KEYWORDS Radiation, citizen science, geographic information systems, exposure assessment