Grantee Research Project Results
2020 Progress Report: Southeast Wisconsin Interdisciplinary Study of Childrens Health, Ecological Exposures and Social Environment
EPA Grant Number: R839278Title: Southeast Wisconsin Interdisciplinary Study of Childrens Health, Ecological Exposures and Social Environment
Investigators: Magzamen, Sheryl , Carter, Ellison , Jathar, Shantanu , Wilson, Ander , Dilworth-Bart, Janean E
Institution: Colorado State University , University of Wisconsin - Madison
EPA Project Officer: Hahn, Intaek
Project Period: January 1, 2018 through December 31, 2020 (Extended to December 31, 2022)
Project Period Covered by this Report: January 1, 2020 through December 31,2020
Project Amount: $600,000
RFA: Using a Total Environment Framework (Built, Natural, Social Environments) to Assess Life-long Health Effects of Chemical Exposures (2017) RFA Text | Recipients Lists
Research Category: Human Health
Objective:
Objective 1: Develop community- and individual-level profiles for social, physical, and chemical environments and determine the relative associations of these exposure profiles with respiratory, neurodevelopmental, and injury-related outcomes in preschool children in Southeast Wisconsin.
Objective 2: Evaluate the role of community-level social and physical environmental profiles on modification of the effect of chemical exposures on children’s respiratory and neurodevelopmental-related outcomes.
Objective 3: Evaluate the role of residential mobility on respiratory, neurodevelopmental, and physical health in preschool children in Southeast Wisconsin.
Progress Summary:
Neighborhood Built Environment Factors
Building on our progress from the prior year, Civil and Environmental Engineering PhD candidate, Oluwatobi Oke, working under the mentorship of project PIs Dr. Carter and Dr. Magzamen, had developed the foundation for the machine learning application for predicting whether houses in our study area contain lead (Project Title: Using physical housing condition to explain variability or prediction in children’s blood lead level in urban areas.)
As part of our EPA project, Mr. Oke’s study proposes the use of machine learning (ML) models to account for some of the limitations listed above. The last decade has marked an increase in the interest of using ML techniques for health research (Lecun et al., 2015; Oskar and Stingone, 2020; Hyejin and Kisok, 2019; Bi et al., 2019; Stingone et al., 2017). ML tends naturally to trend the delineation in large, unprocessed datasets to reveal patterns from datasets, especially knowing the inputs and desired outputs (Cruz et al., 2006). It also allows intelligent systems to build appropriate prediction models and algorithms that classify individual variables with complex interaction factors that might be lacking in other statistical modeling techniques (Galatzer-Levy et al., 2017). The use of the ML approach in this study will allow the use and integration of several variables that could reveal/predict certain patterns in homes with the possibility of childhood lead exposure.
Mr. Oke, under the supervision of Project PI’s Drs. Carter, Magzamen, and Wilson, will use physical housing conditions to predict the presence and dominance source of lead exposure and explain variability in children's blood lead levels between the ages 0-4 years in the city of Milwaukee, Wisconsin, using multiple types of supervised machine learning classification models (e.g. logistic regression, support vector machine (SVM), k-nearest neighbors (k-NN), decision tree and random forest). The study will further explore the association of children with elevated blood lead levels to their neighborhood characteristics using the key characteristics explaining variations in environmental hazards such as income, unemployment rate, educational attainment, health care, and access to transportation.
The outputs from this objective should potentially help to answer the following questions:
(1) Which Pb pathway is likely to have the greatest contribution or dominant source to the high BLL in urban children?
(2) Do environmental profiles interact synergistically to produce comparatively higher BLL outcomes for children with multiple exposures?
(3) Which pathway or route of exposure should be mainly mitigated to reduce the exposure?
Accurate prediction of high-risk residential housings may inform the strategy of decision-makers working to ensure that residents of aging American homes with or without BLL measurements can be identified and as well be able to conduct preventive steps prior to when kids start living in those homes as proposed by the CDC.
Lead Problem in Milwaukee, Wisconsin
According to the US Department of Housing and Urban Development Survey of the prevalence of lead hazard in the US housing, about one-third of housing units in the Midwest have Pb hazards (compared to 25% for the overall United States), and also older housing was more likely to be occupied by families with children (Jacobs et al., 2002). Among the 203,068 children aged under 6 years who were enrolled in Medicaid in 2015, over one-third (n = 71,565; 35.2%) had never been tested for lead. In 2016 alone, 87,443 children under age 6 years received a blood lead test, and 5% (n = 4,353) were identified with elevated blood lead levels. Importantly, the number of children tested has decreased over the past 6 years, with 18,000 fewer children tested in 2016 compared with 2010. Further, the number of children tested in 2016 represents only about 22% of children under age 6 years in Wisconsin, and it is likely that some of the children not tested were at risk for lead exposure and elevated blood lead levels. Thus, the data presented here underestimate the true number of children with elevated blood lead (Christensen et al., 2019). Of the 10 major cities in Wisconsin with at least 100 children tested in 2015, the city of Milwaukee has the highest prevalence rate of elevated blood lead levels at 9.3%. Table 3 shows the demographic composition in the city of Milwaukee.
Future Activities:
Future Research Activities:
- Complete DUA with Wisconsin Department of Health Services for use of STELLAR Lead data, WIC data and Medicaid data.
- Link health data with chemical environment, built environment, and social exposure data.\
- Travel to Milwaukee, WI to conduct Milwaukee Historical Neighborhood tour with David Reimer, former County executive committee.
- Redevelop timeline based on delays in completing DUA with WDHS.
- Resubmit all papers listed under Publications section.
- Develop improved methodological approaches for missing counterfactual population for lead and crime study
- Future Papers
Journal Articles on this Report : 1 Displayed | Download in RIS Format
Other project views: | All 8 publications | 7 publications in selected types | All 6 journal articles |
---|
Type | Citation | ||
---|---|---|---|
|
McGee G, Wilson A, Webster T, Coull B. Bayesian multiple index models for environmental mixtures. BIOMETRICS 2021;1(3). |
R839278 (2020) R835872 (2020) |
Exit Exit |
Supplemental Keywords:
Medicaid, lead, children, injury, respiratory, neurocognitive, coarsened exact matching, machine learning, indoor air quality, residential mobility, WICProgress and Final Reports:
Original AbstractThe perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.