November 17, 2022
Research Highlight

Statistical Learning Shows Silicone Wearables and Urine Samples Reveal Similar Exposure-based Information

Silicone wristbands could be a more convenient option for collecting exposure data

Abstract image of a person wearing a wristband

Researchers use statistical learning to determine similarities in chemical exposure data collected using silicone wristbands and urine. 

(Image: Oregon State University)

The Science                                        

Environmental exposures to chemicals contribute to chronic disease and surveillance approaches use urine samples to characterize an individual’s exposure. In rural areas and during extreme weather events or pandemics, collecting urine samples becomes challenging to assess sensitive populations. An alternative personal chemical exposure monitoring method involves the use of silicone wristbands. The silicone absorbs chemicals from the air and skin, providing information about the wearer’s exposure(s). Cohorts wear the wristbands for several days and then mail them to a laboratory for analysis. Limited information is available on the similarity and accuracy of these monitoring methods such that researchers can use statistical models to predict quantitative information and exposure levels (low, medium, high). In this study, researchers used statistical learning to compare data from urine and wristband samples and determined that it is possible to use computing models to predict exposure levels from inputs of these two distinct data collection modalities.

The Impact

Data collection using silicone wearables than those involving urine samples for measuring personal chemical exposure could increase participation in epidemiological studies and offer a safer method that doesn’t require travel. This study demonstrates that both urine and wristbands are viable methods for gathering similar information which can provide critical information to researchers looking for flexibility to switch between the two collection methods, as needed.


Many studies use urine samples to measure chemical exposure in humans. Based on data from the samples, researchers use computational models to predict measurements and exposure levels (low, medium, and high). Collecting urine samples, however, can be challenging for those in rural areas and during extreme weather events or pandemics. A new sampling alternative, silicone wristbands or wearables, allows for continuous collection during such events. The wristbands absorb chemicals around a person, which can be easily mailed to the laboratory where researchers can determine what the person was exposed to, as well as the exposure amount. The purpose of this study was to compare chemical exposure data gleaned from urine samples and wearables to determine if they’re similar enough for researchers to apply the same statistical models on both datasets.

PNNL Contact

Lisa Bramer, Pacific Northwest National Laboratory,


Research reported in this publication was supported by the National Institute of Health (NIH) under award number UH3OD023290 and the National Institute of Environmental Health Sciences (NIEHS) under award numbers 1R21ES024718, 4R33ES024718, P30ES030287, and P42ES016465. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH or NIEHS. Holly M. Dixon was supported in part by NIEHS Fellowship T32ES007060 and the ARCS Foundation®. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control (CDC). Use of trade names is for identification only and does not imply endorsement by the CDC, the Public Health Service, or the U.S. Department of Health and Human Sciences.

Published: November 17, 2022

Dixon H.M., L.M. Bramer, R.P. Scott, L.P. Calero, D. Holmes, E.A. Gibson, and H.M. Cavalier, et al. 2022. "Evaluating predictive relationships between wristbands and urine for assessment of personal PAH exposure." Environment International, 163, PNNL-SA-163167. [DOI: 10.1016/j.envint.2022.107226]