A Machine Learning Based Approach for Estimating Specific Gravity in Elementary-School Aged Children
23 Pages Posted: 21 Sep 2022
Abstract
Environmental research often relies on urinary biomarkers which require dilution correction to accurately measure exposures. Specific gravity (SG) and creatinine (UCr) are commonly measured urinary dilution factors. Epidemiologic studies may assess only one of these measures, making it difficult to pool studies that may otherwise combine.Participants from the National Health and Nutrition Examination Survey 2007-2008 cycle were used to perform k-fold validation of a nonlinear model estimating SG from UCr. The final estimated model was applied to participants from the School Inner-City Asthma Intervention Study, who submitted urinary samples to the Children’s Health Exposure Analysis Resource. Model performance was evaluated using calibration metrics to determine how closely the average estimated SG was to the measured SG. Additional models, with interaction terms for age, sex, body mass index, race/ethnicity, and relative time of day when sample was collected were estimated and assessed for improvement. The association between MBZP and asthma symptom days, controlling for measured UCr, measured SG, and each estimated SG were compared to assess validity of the estimated SG.The model estimating SG from UCr alone, resulted in a beta estimate of 1.11 (95% CI: 1.03, 1.20), indicating agreement between model-predicted SG and measured SG. The full model accounting for all interaction terms with UCr resulted in the best agreement (β= 1.05, 95% CI: 0.97,1.13). Our nonlinear modeling provides opportunities to estimate SG in studies that measure UCr or vice versa, enabling data pooling despite differences in urine dilution factors.
Note:
Funding Information: This work was supported in part by funding from NIH/NIEHS: U2CES026561, U2CES026553, U2CES026555, R00ES027508, R01AI073964, R01AI073964-02S1, K24AI106822, K23ES031663, U01AI110397, and P30ES000002. Dr. Hauptman was also supported by the American Academy of Pediatrics (AAP) and funded in part by cooperative agreement award with the Centers for Disease Control and Prevention/Agency for Toxic Substances and Disease Registry (CDC/ATSDR) FAIN:NU61TS000296. The U.S. Environmental Protection Agency (U.S.EPA) supports the Pediatric Environmental Health Specialty Units (PEHSUs) by providing partial funding to the ATSDR under Inter-Agency Agreement DW-75-95877701.
Declaration of Interests: The authors declare that they have no competing interests.
Ethics Approval Statement: Data for this project was obtained from the publicly available data in the Human Health Exposure Resource (HHEAR) Data Repository, which has been approved under Icahn School of Medicine at Mount Sinai IRB Protocol # 16-00947. The National center for Health Statistics Research Ethics Review Board approved documented consent for all NHANES participants.
Keywords: Keywords: dilution factors, data pooling, calibration metrics
Suggested Citation: Suggested Citation