&EPA
United States
Environmental Protection
Agency
Computational Exposure Science at the
Environmental Protection Agency's
Office of Research & Development
Presentation to the American Chemistry Council's Long Range
Initiative Strategic Science Team
Peter P. Egeghy, Ph.D., M.P.H.
Computational Exposure and Toxicokinetics Branch%
Chemical Characterization and Exposure Division
2 June 2020
Office of Research and Development
Chemical Safety for Sustainability Research Program
Disclaimer: The views expressed in
this presentation are those of the
author and do not reflect the
views or policies of the U.S.
Environmental Protection Agency.
-------
xvEPA
United States
Environmental Protection
Agency
Evolving Emp
Exposure Science
TT21: Exposure must be integrated with
toxicity testing at every step of risk
assessment to guide development and
use of toxicity information
Predictive models are needed to screen
chemicals based on exposure
ExpoCast created to complement ToxCast
by building on the technological and
computational advances in exposure science
Our goal has been to advance the
characterization of exposure to inform
chemical prioritization for evaluation as well
as to translate results of high-throughput
toxicity testing
Office of Research and Development
Chemical Safety for Sustainability Research Program
Slide 2
E
tho 21st Century
¦m ANO A 5TRATF3V
TOXICITY TESTING IN THE 21 ST CENT
A VISION AND A STRATEGY
- ^ v>* > ¦>*:
-------
vvEPA Transition to 21st
United States
Environmental Protection
Agency
Hazard f
Dose-Response
Relationships
Identification
Exposure
Assessment
NCER
NHEERL/
NCCT
NERL
Risk
Characterization
Risk
Management
NCEA
NRMRL
Office of Research and Development
Chemical Safety for Sustainability Research Program
Chemical
Toxicity
Toxicity
Extrapolation Modeling
CEMM
CPHEA
Slide 3
https://www.epa.qov/sciencematters/epas-office-research-and-development-reorqanizes-better-support-epas-mission
-------
<>EPA Center for Computati
Environmental Protection
Exposure: In
CCTE works to support Agency decisions by providing solutions-
driven research to rapidly evaluate the potential human health and
environmental risks. CCTE research strives to:
Reduce the time required to thoroughly test chemicals and
emerging materials for human and ecological toxicity from years
to months.
Expand our understanding of quantitative human and ecological
exposures for thousands of chemicals and emerging materials.
Develop a comprehensive system of actionable chemical safety
and ecological data with the software tools to integrate them for
a range of human health and environmental decisions.
Demonstrate translation of CCTE data, models, and tools into
regulatory decisions by EPA Program Offices, EPA Regions, and
States to protect human health and the environment.
Using the knowledge and tools developed from this research, CCTE
performs rapid chemical screening and evaluation that allows
thousands of chemicals to be evaluated for potential risk in a very
short amount of time. The data and tools produced by CCTE
researchers are intended to help Region and Program Offices, states,
tribes, and communities make decisions to sustain a healthy society
and environment.
https://www.epa.aov/aboutepa/about-center-computational-toxicoloav-and-exposure-ccte
Office of Research and Development
Chemical Safety for Sustainability Research Program
Center for Computational Toxicology and Exposure (CCTE)
- Evaluates chemical toxicity through a variety of novel
methods, including HTT, AOPs, VTM, and ETAM
- Models chemical exposure (Rapid Exposure Modeling
and Dosimetry, ExpoCast) to contextualize hazard
- Disseminates chemical toxicity and exposure data and
predictive tools (e.g., through the "CompTox Chemistry
Dashboard")
CCTE has Four Divisions
- Biomolecular & Computational Toxicology Division
- Chemical Characterization & Exposure Division
- Great Lakes Toxicology & Ecology Division
- Scientific Computing & Data Curation Division
Slide 4
-------
oEPA Chemical Characterization and
United States
Environmental Protection , m m
Exposure D
CCED) performs research to develop and advance
experimental chemistry approaches that are critical to the
rapid characterization of the presence, structural
characteristics, and properties of chemicals that are of
interest to EPA scientists due to their potential
environmental fate and toxicity. In addition to chemical
characterization, CCED develops computational models to
predict external exposure and internal doses for large
numbers of chemicals based on minimal data.
CCED strives to reduce the time to conduct toxicity and
exposure assessments from years to months by
developing:
Chemoinformatic tools and knowledgebases
Rapid analytical methods for identifying environmental
chemicals in environmental and biological samples
Predictive models of both exposure and dose for
environmental chemicals
Absorption, Distribution, Metabolism and Excretion
approaches for environmental chemicals and model
parameterization
Examples of Research in CCED:
ExpoCast
HTTKR Package
Non-Targeted Analysis Collaborative Trial (ENTACT)
CompTox Chemicals Dashboard
Chemical and Products Database (CPDat)
DSSTox
Toxicity Estimation Software Tool
Adverse Outcome Pathway
Predicting screening-level population exposure and intake
dose rates by strengthening linkages from structure, to
function, to use scenarios, to dose by combining
information on:
Chemical properties
Product formulations
Mechanistic fate and transport processes
Consumer behavior informatics
Improved methods for extrapolating across chemicals
https://www.epa.qov/aboutepa/about-chemical-characterization-and-exposure-division
I Office of Research and Development
I Chemical Safety for Sustainability Research Program
Slide 5
-------
xvEPA
United States
Environmental Protection
Agency
Computational Exposure and
Toxicokinetics Branch
Jeff Daniel
Minucci Dawson
Research Research
Triangle Park, NC Triangle Park, NC
Kristin
Isaacs
Research
Triangle Park, NC
Peter
Rogelio
Tornero-Velez
Research
Triangle Park, NC
Katherine
Phillips
Research
Triangle Park, NC
John
Wambaugh
Research
Triangle Park, NC
A FA
Marina Hisham
Evans El-Masri
Research Research
Triangle Park, NC Triangle Park, NC
Elaina
Kenyon
Research
Triangle Park, NC
Chief
Research
Triangle Park, NC
-------
&EPA
United States
Environmental Protection
Agency
Computational Ex
Integrating Data Streams and Models
Functional
role
Formulation
Science
1
Product
formulation
J.
t
Product
purchase
Product
use
Behavior
Informatics
I
Chemical
release
I
Media
concentration
Exposure
and Dose
Modeling
I
Exposure
I
Dose
Inherent Chemical
Properties
Human Decisions
and Behavior
J
Slide 7
I
High-throughput Predictions of
Population-Level Chemical Exposures and Intake
Egeghy et al., EHP, 2016
-------
xvEPA
United States
Environmental Protection
Agency
Computational Ex
Integrating Data Streams and Models
Functional
role
Formulation
Science
I
Product
formulation
I
Product
purchase
Product
use
Behavior
Informatics
I
Chemical
release
I
Media
concentration
I
Exposure
and Dose
Modeling
Exposure
I
Dose
1 n h e re ri t Chemical
Properties
Human Decisions
and Behavior
J
Datasets
Chemical molecular structures and
chemical function
Chemical end-use including presence
in consumer products
Population patterns of consumer
product purchasing
Patterns of consumer product
use
Properties of homes and indoor
environments
Chemicals measured in residential
media
Human activity patterns and
exposure factors
Human physiology and
pharmacokinetic properties of
chemicals
Slide 8
t
High-throughput Predictions of
Population-Level Chemical Exposures and Intake
-------
&EPA
United States
Environmental Protection
Agency
Computational Ex
Integrating Data Streams and Models
Models predicting
chemical function from
chemical properties or
structures
Formulation
Science
Behavior
Informatics
Models describing fate
and transport of
chemicals within a
residence
Exposure
and Dose
Modeling
Probabilistic human
exposure models
Dosimetry/
pharmacokinetic models
Inherent Chemical
Properties
Human Decisions
and Behavior
Functional
role
Dose
Product
formulation
Product
purchase
Product
use
Chemical
release
Media
concentration
Exposure
I
Slide 9
High-throughput Predictions of
Population-Level Chemical Exposures and Intake
-------
&EPA
United States
Environmental Protection
Agency
The Chemicals and Products Database
(CPDat)
Reported
Chemicals in
Products
Chemical
Role in
Products
CPCat
CPCPdb
Ingredient
Lists
Chemica and Products Database
Functional
Use Data
Measured
Data
Broad Chemical
Categories
Quantitative
Identification of
Compounds
Office of Research and Development
Chemical Safety for Sustainability Research Program
Slide 10
-60,000 chemicals
~16,000 products
-300 consumer product
categories
Dionisio et al., Sci Data, 2018
Isaacs et al., JESEE, 2018
Phillips et al., Green Chem, 2017
Phillips et al., ES&T, 2018
-------
xvEPA
United States
Environmental Protection
Agency
ChemExpoDB/Factotum
Factotum- web interface for exposure data curation
ChemExpoDb - integrated family of databases to hold exposure data (use,
monitoring data, product information, toxicokinetic data)
Internal (and eventually external) webservices are being built to provide data
in a machine-readable form to the CompTox Chemicals Dashboard and
stakeholders
PDF -> text
Data extraction
¦=>
Data loading
Multiple data
streams
¦=>
Data harmonization
(common terminology &
units)
Assignment to common
organizational schema
Data curation
J' Welcome to Factotum
Documents Products
511,230 586,153
Data analysis
Inputs for modeling
Populate CompTox
Chemicals Dashboard
Factotum
Products Linked To
PUCs
60,624
Curated Chemical
Records
1.9 million
Extracted Chemicals
3.9 million <
Unique DTXSIDs
28,243
Web services
Office of Research and Development
Chemical Safety for Sustainability Research Program
Slide 11
ChemExpoDb
Courtesy of K, Dionisio
-------
Refining
Environmental Protection * ¦ *"
Agency
- Developed partnership with retailers and national marketing
research companies to obtain geographically-specific data on
purchasing habits (use surrogate), including household-level
purchasing frequency data
- Refinement of consumer product categories
- Identifying and incorporating other available sources of consumer
product use data
National Institute of Environmental Health Sciences' Sister Study
Small-scale studies
- Still exploring use of innovative web-based infoveillance methods
F/i SISTER STUDY | idSCH (lO /QIC
I / } A Study of the Environmental and Genetic Risk Factors for breast cancer M. m m m M W Jr
NATIONAL INSTITUTE OF ENVIRONMENTAL HEALTH SCIENCES NATIONAL INSTITUTES Of HEALTH DEPARTMENT Of HEALTH AND HUMAN SERVICES ~ jg
Office of Research and Development
Chemical Safety for Sustainability Research Program
Slide 12 Courtesy of T. Buckley
-------
Methods
Collaboration with the Nielsen
HomeScan Panel
¦ Resulting Data and Ongoing Analyses
Longitudinal data for 2012 for 60,000 U.S. households
>4 million individual purchases of 200 product types
Understanding unique patterns of product purchases: demographic or geographic
patterns of high consumer product use or co-occurrence patterns of products
Merging databases of consumer product ingredients with product purchase:
ultimately identifying chemical co-occurrence patterns and potential cumulative
exposures
I Office of Research and Development
I Chemical Safety for Sustainability Research Program
nielsen
4>EPA Consumer
United States
Environmental Protection A _ _ A
Agecv uatasets
Insecticide- Ant Killer (Liquid)
¦ Research Question: What chemical-
containing products are U.S. households
purchasing and in what amounts and how
often?
Slide 13
Courtesy of K. Isaacs
-------
xvEPA
United States
Environmental Protection
Agency
Market Basket
Product Purchase Data
Product Category
re
-Q
o
u
/
/
a#
f
/
Detergents
1
0
0
0
0
Vitamins
2
0
0
-4
-3
Hair Care
3
0
0
-6
1
Personal Soap and Bath
4
0
-2
2
1
Laundry Supplies
5
-3
1
2
1
Household Cleaners
6
-1
1
2
0
Oral Hygiene
7
1
0
2
0
Pet Care
8
-2
0
-7
-3
Skin Care Preps
9
4
0
1
1
Fresheners and Deodorizers
10
-2
0
3
1
Deodorant
11
-2
0
1
1
Household Supplies
12
1
0
1
Cosmetics
13
4
0
1
1
Paper Products
14
0
0
1
1
Insecticides
15
-2
-1
1
-1
Stationary & School Supplies
16
1
1
-3
1
Automotive
17
-1
0
-1
-1
Shaving Needs
18
-1
0
-5
1
Medications, Remedies
19
3
0
-1
0
First Aid
20
-2
-1
3
0
Floral Gardening
21
1
1
-3
-2
Tobacco & Accessories
22
-2
0
-3
-2
Baby Needs
23
2
0
2
2
Fragrances- Women
24
1
0
2
2
Men's Toletries
25
0
0
-1
0
Ethnic HABA
26
-3
-3
10
-1
Hardware, Tools
27
1
1
-1
-1
Feminine Hygiene
28
1
1
1
2
Elec,Rec,Tape
29
1
1
0
0
/ / / / ^
P -o »o -o
-------
xvEPA
United States
Environmental Protection
Agency
Investigate Produ
Data for Co-occurring EDCs
Co-occurring Endocrine Disrupting Chemicals Rank by Demographic Group
Top 20 chemical sets occur in ~500
or more household-months
Cooccurrence example:
2-hydroxy-4-
methoxybenzophenone
used in sunscreens,
widespread in things like
plastics and toys
FD&C blue no, 1
used in children's medications,
cosmetics
Higher preference: households
with children; woman of
childbearing age
Lower preference: African
American households
Demographic
0 {l-cedr-8-en-9-yletiianone,2-hydroxy-4-methoxybenzophenone}
l l {diazolidinyl urea,2-hydroxy-4-methoxybenzophenone}
2 {2-hydroxy-4-methoxybenzophenone,benzethonium chloride}
0 {propylparaben,fd&c blue no, 1}
1 {l-cedr-8-en-9-ylethanone.fd&c blue no. 1}
0 {diazolidinyl urea, propylparaben)
1 {propylparaben.benzethonium chloride}
|2-hydroxy-4-methoxybenzophenone, propylparaben)
-3 {2-hydroxy-4-methoxybenzophenone,rd&c blue
0 -2 {2-hydroxy-4-methoxybenzophenone,propylparaben.td&c blue no. 1}
¦ n Demographic
P Race/Ethnicity
Education
.40 ¦lncome
[_ Family Comp.
. -60 Female Age
-14 - 74 -15 -27 j -19 0 1 {1-C
-8 11 {2-hydroxy-4-methoxybenzophenone,dimethyldiactadecylammonium chloride}
-7 0 0 {propylparaben.dimethyldioctadecylammonium chloride}
-2 11 {dl-tocopherol mixture,phytonadione}
14 1 1 {dl-tocopherol mixture,propylparaben}
10 0 {l-cedr-8-en-9-ylethanone.propylparaben}
-6 {dl-tocopherol mixture,2-hydroxy-4-methoxybenzophenone}
0 {1-tetradecanamine. n.n-dimethyl-, n-oxide,l-cedr-8-en-9-ylethanone}
0 {1-tetradecanamine, n.n-dimethyl-, n-oxide,propylparaben}
0 {1-tetradecanamine, n.n-dimethyl-, n-oxide,2-hydroxy-4-methoxybenzophenone}
-cedr-8-en-9-ylethanone,dl-tocopherol mixture}
hair conditioner
g
haii
i
sur
I
pre:
,
unk
I
mas
I
UV absorbe
0
I
colorant
~ °
fragrance
0
.
masking agent
-lo
b
Office of Research and Development
Chemical Safety for Sustainability Research Program
Slide 15
Courtesy of K. Isaacs
-------
Quantitative
Relationships (QSURs)
Environmental Protection
Agency
¦catalyst!
croMlinkci
[emulsifier]
I film I
I forming
| agent |
|pH itabiluer|
Irheology I
| modifier |
|pres«rvatlva|
I wetting I
I Standard Deviation Hi Y-randomization Error
I Missciassification Error f J 5-fold Cross Validation Error
Colorants
Suite of QSUR Models
Chemicals that have
no reported use
Predicted Uses for Chemicals
Catalysts
Crosslinkers
Office of Research and Development
Chemical Safety for Sustainability Research Program
Slide 16
Phillips et a\., Green Chem., 2017
-------
xvEPA
United States
Environmental Protection
Agency
Applications
Consumer Product
Chemical
Compositions
Food and Drinking
consumer
Product Use
Patterns Concentrations
Scenarios File
(Scenarios/Routes
Parameterized by / r.
Chemical) ^ J
Chemical Ij,
Properties
o
SHEDS-HT
Probabilistic
Exposure
Model
Indoor
Fugacity
Module
Lr
Indirect
Exposure
Module
Population
Module
-- t
Direct
Exposure
Modules,
s.
Dietary
Exposure
. Module,
Exposure Aggregation, Dermal
, Removal Processes, and Intake Dose
Exposure Factors
1e
Distributions of
Predicted Exposures
and Intake Doses
Isaacs et al., Env. Sci. & Tech., 2014
aSS' i '..¦₯> . 1. * k ¦"
¦Vl"'" 111 I ¦
m wlx .
psapipj}
f i s WmFTOf
I w Y
Isaacs et al.. Tox. Rep.. 201
Exposure Estimate Parameterization
Functional Substitutes
*| t
- I1 i
2 i i
Functional Use
Threshold
Bioactivit^
Candidate Alternatives
.
Phillips et a I., Green Chem.. 2017
Alternatives Screening
¦ >¦ ¦ ¦¦¦ ¦¦¦¦
¦¦ ¦¦¦¦¦ ¦¦¦¦¦¦¦
m ii ¦¦¦¦in
¦¦¦ ¦ mmimi ii
in ¦ ¦ ¦ ¦
¦ ¦ ¦ ¦¦¦¦¦ ¦¦¦¦¦¦¦¦¦¦
¦ ¦ ¦¦¦¦¦¦¦¦¦¦¦ mil
II ai mil
¦ ¦¦ ¦¦¦¦¦¦>¦
¦ I III mi
Reported Chemical Function
mi ii in
III Hill
III una
Unique Chemicals
Predicted Functions
Phillips et al., Environ. Sci. & Tech., 2018
Suspect Screening Identification
Office of Research and Development
Chemical Safety for Sustainability Research Program
Slide 17
Courtesy of K. Phillips
-------
&EPA
United States
Environmental Protection
Agency
SHEDS-High Throughput:
Merging Multiple Data Streams
Consumer Product
Chemical Composition
Chemical Concentrations
in Food and Drinking
Water
/
\
Consumer
Product Use
Patterns
V
J
T>
Chemical-specific list
of all parameterized
exposure scenarios/
routes
Structured "Scenarios File'
SHEDS-HT
Probabilistic
Exposure
Model
Fugacity-Based
Source-to-
Concentration
Module
Indirect
Exposure
Module
Dietary
Exposure
Module
Updated as more information
becomes available
Census Data
Human Activities
Exposure Factors
Food and
Water Intake
Data
Office of Research and Development
Chemical Safety for Sustainability Research Program
Input
Databases
Census
Fluman Activity
Ambient Cone.
Food Residues
Recipe/Food Diary
Exposure Factor
Distributions
~ ~
Calculate Individual
Exposure/Dose Profile
Output
' Population Exposure
f
¦ Population Dose
o
?>
Source: H. Ozkaynak
- Reduced version of SHEDS-Multimedia;
stochastic methods
- Available as R package
- Inputs
Use information and scenario mapping
Fugacity modeling methods used to
determine air and surface concentrations
for near-field indirect scenarios
- Outputs: Exposures for key cohort groups, by
pathway(m g/kg/day)
- Updated to handle articles and dietary
exposures
- Recently underwent external peer review
- Active collaboration with industry and academic
partners on dietary and consumer chemicals
Slide 18
https://qithub.com/HumanExposure/SHEDSHTRPackaqe
Courtesy of K. Isaacs
-------
xvEPA
United States
Environmental Protection
Agency
Moving Toward
Higher-Tier A
Chemical Human Exposure Model
9 0
* c
m
S?
1 -MSL
Population
Characteristics
V
i I wf
i it;
Residential
Characteristics
Behavior and
Product Use
~
Source-to-Dose
(based on SHEDS- I
HT)
}
Chemical Exposure
(days to years)
eOlSlOTf;
Product
Composition
(CPDat)
Chemistry
Dashboard
Office of Research and Development
Chemical Safety for Sustainability Research Program
Slide 19
Courtesy of K. Isaacs
-------
xvEPA
United States
Environmental Protection
Agency
Understanding & Predicting Chemicai
Occurrence in Environmental Media
Large database of multimedia
monitoring information (~200
million records) obtained from 21
public databases
Harmonized to chemical identifier
(DTXSID) and 32 unique media
Will allow for more efficient and
rapid identification of available
monitoring data for chemicals of
interest
Will form basis for machine-
learning models of occurrence in
media for use in non-targeted
workflows and screening-level
assessments
Office of Research and Development
Chemical Safety for Sustainability Research Program
Slide 20
Courtesy of K. Isaacs
Medium
ambient air
breast milk
drinking water
food product
groundwater
landfill leachate
human (other tissues or fluids)
human blood (whole/serum/plasma)
indoor air
indoor dust
iivestock/'meat
personal air
product
raw agricultural commodit
other-ecological
other-environmental
precipitation
raw agricultural commodil
sediment
skin wipes
sludge
soil
surface water
unknown
urine
vegetation
wastewater (influent, efflu
wildlife (aquatic invertebrs
wildlife (aquatic vertebrate
wildlife (birds)
wildlife (fish)
wildlife (terrestrial invertet
wildlife (terrestrial vertebrates)
Unique Chemicals
581
Log(» Samples)
4.5-5.0
: c--j :
5.M.0
n - o ¦
6.5-7 0
7 0-7.5
Log(» Samples)
0.0-1.0
1.0-2 0
2 0-3.0
3.0-4,0
4.0-50
25.0-60
60-7.0
70-6.0
-------
&EPA
United States
Environmental Protection
Agency
High Throughput Toxi
In vitro toxicokinetic data
In vitro
hepatic
clearanc
e
Plasma
protein
binding
^ v ^ v ^ v # _3
o
I
*
IS
2*
V
I't
mm
Administrator Wheeler (September, 2019):
"I am directing leadership and staff in the
Office of Chemical Safety and Pollution
Prevention and the Office of Research and
Development to prioritize
...the reduction of animal testing while
ensuring protection of human health and the
environment."
Office of Research and Development
Chemical Safety for Sustainability Research Program
generic toxicokinetic model
Kidney Tissue
Kidney Blond
^^^kidne^
Gut Lumen
Gut BlooH~
Liver Tissue
Liver mood"
Rest of Body
Body BloocT"
httk
high(er) throughput toxicokinetics
Slide 21
Courtesy of J. Wambaugh
-------
oEPA
sspj, p^ion Building Conf
_Q
"D
X
Chemical-specific TK model allows comparison of predictions
to in vivo data
¦ Can estimate bias and uncertainty
¦ Can extrapolate to other situations (dose, route,
physiology) where you don't have data
As most chemicals lack chemical-specific data, we need a
generic TK model
¦ Expect larger uncertainty, but also greater confidence in
model implementation
¦ Can estimate bias and uncertainty, and try to correlate with
chemical-specific properties
¦ Can use model to extrapolate to other situations (chemicals
without in vivo data)
Constructing an in vivo blood/plasma/tissue concentration vs.
time (CvT) database to evaluate high throughput PBTK models
for chemical prioritization and regulatory decision making
>
o
CO
c
o
-*>
03
-I'
c
0
o
c
o
o
_Q
"O
I
>
o
CO
c
o
-*>
03
-I
c
0
o
c
o
o
X/
A
X/
X/
/X
'' X
Chemical
Specific
Model
Predicted Concentrations
x
y/
y y
x Z
x/' z z
y / x
X y V
y z Generic
/' x z HTTK
'' y Model
7
Predicted Concentrations
Office of Research and Development
Chemical Safety for Sustainability Research Program
Cohen Hubal et al., JESEE, 2019
Slide 22
-------
oEPA
United States Q| f av|fy|^fq/
Environmental Protection C# f f f f f f Clf w
Agency J
Exposure science has a new setting in the Office of Research &
Development
- No longer isolated, now better integrated with toxicity testing
- Follows the evolution of chemical evaluation
The Center for Computational Toxicology and Exposure
- Exploits advances in technology
-Aims to be able to rapidly evaluate thousands of chemicals
- Provides contextualization of high-throughput toxicology
- Is problem-driven and solution-focused
Broad applications beyond the Agency
- Partnerships with Minnesota Dept. of Health and California Dept of
Toxic Substances Control as examples
Office of Research and Development
Chemical Safety for Sustainability Research Program
Slide 23
-------
£EPA
United States
Environmental Protection
Agency
Acknowledgements
Kristin Isaacs
John Wambaugh
Katherine Phillips
Rogelio Tornero-Velez
Kathie Dionisio
Zach Stanfield
Jeffrey Minucci
Daniel Dawson
Annette Guiseppi-Elie
Michael-Rock Goldsmith
Timothy Buckley
Linda Sheldon
Elaine Cohen Hubal
Office of Research and Development
Chemical Safety for Sustainability Research Program
-------
xvEPA
United States
Environmental Protection
Agency
Occupational Expos
for Exposure-based Prioritization
Need to parameterize for 1000s of
substances lacking data
Using existing workplace exposure
data to develop a model that can
predict air concentration based on
chemical/physical properties and
industry type
Using Bayesian hierarchical logistic-
regression
- First predict detect/nondetect
- Next predict concentration
Preliminary results look promising
Currently predicting NAICS sector/
subsectors with functional use models
Furniture and Related Product Manufacturing
Leather and Allied Product Manufacturing
Personal and Laundry Services
Printing and Related Support Activities
Chemical Manufacturing
Apparel Manufacturing
Specialty Trade Contractors
Religious, Grantmaking, Civic, Professional, and Similar Organizations
Miscellaneous Manufacturing
Textile Mills
Animal Production and Aquaculture
Fabricated Metal Product Manufacturing
Paper Manufacturing
Repair and Maintenance
Nonstore Retailers
Real Estate
Nonmetallic Mineral Product Manufacturing
Wood Product Manufacturing
Computer and Electronic Product Manufacturing
Administration of Environmental Quality Programs
Motor Vehicle and Parts Dealers
Transportation Equipment Manufacturing
Textile Product Mills
Beverage and Tobacco Product Manufacturing
Support Activities for Agncuiture and Forestry
Electrical Equipment, Appliance, and Component Manufacturing
Food Manufacturing
Sporting Goods, Hobby, Musical Instrument, and Book Stores
Primary Metal Manufacturing
Machinery Manufacturing
Hospitals
Educational Services
National Security and International Affairs
Merchant Wholesalers, Durable Goods
Ambulatory Health Care Services
Museums, Historical Sites, and Similar Institutions
Professional, Scientific, and Technical Services
Rental and Leasing Services
Furniture and Home Furnishings Stores
Building Material and Garden Equipment and Supplies Dealers
Support Activities for Transportation
Petroleum and Coal Products Manufacturing
Plastics and Rubber Products Manufacturing
Water Transportation
Nursing and Residential Care Facilities
Gasoline Stations
in of Housing Programs, Urban Planning, and Community Development
Probability of Detection in Air
Test set
£
cii
E
?
Office of Research and Development
Chemical Safety for Sustainability Research Program
Slide 25
<
r= 0.79
-12.5 10.0 -7.5 5.0 2.5 0.0 2 5 5 0 7 5
Predicted air concentration (log mg/m3)
------- |