www.epa.gov/research Technological Tools for Evidence Integration Shane Thacker, Jennifer Nichols, Ryan Jones, Steven J. Dutton National Center for Environmental Assessment; Office of Research and Development; US EPA Shane Thacker I thacker.samuel@epa.gov I 919-541-5159 Evidence Integration New Tools, continued New Tools, continued At the EPA's National Center for Environmental Assessment (NCEA), we work closely with programs throughout the EPA to integrate web-based and desktop computer tools into the assessment process, facilitating evidence integration for science assessment products. By incorporating in-house and third-party tools, both open source and commercial, activities such as the Integrated Risk Information System (IRIS) and the Integrated Science Assessments (ISA) seek to use the best tools for the job, while remaining flexible enough to improve the evidence integration process. Current Tools • Health and Environmental Research Online (HERO): Literature search, categorization, acquisition, archiving. Interoperable with HAWC, Distiller, and SWIFT. • Health Assessment Workplace Collaborative (HAWC): Study evaluation, data extraction, visualization. Interoperable with HERO, BMDS, and Distiller. • Benchmark Dose Software (BMDS): Dose-response modelling. Interoperable with HAWC. • Evidence Partners DistillerSR: Literature screening, data extraction. Interoperable with HERO and HAWC. • Sciome SWIFT-Review and SWIFT-Active Sereener: Literature screening, prioritization, categorization. Interoperable with HERO and HAWC. Evidence Profile Table, pictured New Tools Evidence Profile Table Part of HAWC, the Evidence Profile Table offers a summary explanation of evidence integration in a chemical risk assessment. This view creates greater transparency about the body of evidence by illuminating the rationale behind the assessment findings. • Adaptation of GRADEPro Evidence Profiles • Create multiple rows to cover multiple evidence streams • Select studies and endpoints added to HAWC • Streams break down into scenarios • Endpoints are rated within scenarios • Confidence judgements build from individual to across-stream • Findings summaries add to confidence judgments * Profile Streams Scenario Studies (organized by effect tag) Factors That Increase Confidence In The Selected Studies Profile Stream Scenarios Factors That Decrease Confidence In The Selected Studies Outcome Studies Factors That Increase Confidence Factors That Decrease Confidence Confidence Judgement and Summary of Findings for Individual Outcome Within- Stream Confidence Judgement and Summary of Findings Inference Across Streams Across- Stream Confidence Judgement test animal (Animal Bioassay) •test inference this is another description ®©o test total This is a test explanation Test scenario neoplastic •Initial submission: Title of the study (Thacker and Jones 2019) •Consistency Relatively Low •High Risk of Bias This is a factor that would decrease confidence. ©fflO Moderate test outcome This is another explanation eoo Slight test judgement This is a full explanation test outcome Overall Odds Ratio • Title of another study here (Thacker and Nichols 2019) •Consistency Relatively Low •High Risk of Bias This is another factor that would decrease confidence. ©e® Robust test title test test summary within test summary full Title of summary And the full summary. Evidence Mapping Integrated with the Health and Environmental Research Online (HERO) database, the Evidence Mapping tool allows researchers to create heat maps to visualize and overlay characteristics (e.g., discipline, exposure, concentration, etc.) of the reviewed literature, making it easy to visualize the available evidence. HERO tags used for Evidence Map Example Evidence Map a Health Outcome Category Mortality CV 14 Resp 12 Repro/ Nervous System Cancer Dermal i t US 16 11 2 0 0 0 5 128 Canada 10 1 1 0 0 3 7 48 6 2 5 0 i 5 116 Asia 5 3 8 1 10 178 Other ii^B ¦» 2 1 0 0 2 128 E to 8? US 15 8 18 3 5 5 0 7 106 Canada 5 1 4 1 1 0 : 17 Europe 5 5 11 16 6 2 0 1 46 Asia 10 16 19 25 4 4 1 6 89 Other 3 12 3 15 0 0 4 42 1 US 3 1 3 1 1 1 1 1 13 Canada 0 0 1 0 0 0 0 0 C 1 Europe 0 3 4 2 1 0 0 0 2 12 Asia 0 1 8 2 0 0 0 2 0 13 Other 0 1 2 2 0 4 0 1 1 11 Total 140 227 310 135 27 32 15 10 52 948 • In HERO, scientists use tags to categorize literature for possible use in chemical risk assessment projects • Using the tool, scientists create crosstabs between sets of tags that code literature by characteristics • The results are color-coded, creating heat maps for easy visualization of the intersection totals • The result is a map showing the amount of possible evidence between characteristics, such as location and exposure • Researchers can layer the characteristics into sets and subsets, adding visual organization • Interoperable with Distiller Evidence Inventory The Evidence Inventory tool, hosted within HERO, facilitates data extraction and portrayal by providing researchers a template to collect and categorize data from the relevant literature and then create summary tables of the extracted information. The summary tables are then ready for export into assessment documents, allowing readers to review the evidence behind the chemical risk assessment. ,00 Reliability/St Study Study Sample HERO ID Author Year Title ISA Relevant Relevant udv Quality Section Health Endpoint Design Name Study Population Details Size Com 123457 Thacker, S;Joi 2015 123457 Thacker, S; loi 2015 123457 Thacker. 5; Jo>2015 123459 lo 123459 Jo , R; Dutt 2014 , R; Dun 2014 High Quality High Quality High Qu; High Quality High Quality High Quality High Qi >i Yes gasol Yes gasol Yes gasol Yes gasol Yes High Qu; High Quality High Qui Infl/Ox Stres Infl/Ox Sti Infl/Ox Sti • Producing a chemical risk assessment document requires extracting and reformatting the data in cited studies into tables • Previously, this was largely a manual task without required standardization • This new tool uses spreadsheets formatted for different disciplines to help standardize data extraction • Once the data is extracted, the spreadsheets are transformed into sets of tables useful in the document production process Future Development • Store data in HERO for repeated use in assessments • Develop search and reporting capabilities for extracted data • Through text and concept mining tools, automate the first pass at categorization and tagging • Visualize the results of automated categorization in Evidence Maps • Create tighter integration between HERO and HAWC • Create, improve, and utilize web service APIs for HERO and HAWC to ease integration with third-party software • Allow web-based data entry for Evidence Inventory • Integrate Tableau visualization software with HAWC, Distiller, and Evidence Inventory tools • Integrate Evidence Prime's Pupil automated data extraction software with Distiller, HAWC, and Evidence Inventory tools • Investigate possible standards for extracted data formats to ease data migrations • Investigate and implement tools for automated table and graph data extraction • Work on ontologies for data extraction to make the data more easily searchable • Employ agile development processes to test and incorporate new and useful tools into the assessment process U.S. Environmental Protection Agency Office of Research and Development Disclaimer: The views expressed in this poster are those of the authors and do not necessarily represent the views or policies of the U.S. Environmental Protection Agency. Any mention of trade names, products, or services does not imply an endorsement by the U.S. Government or the U.S. Environmental Protection Agency (EPA). The EPA does not endorse any commercial products, services, or enterprises. ------- |