An Environmental Heath Vocabulary and its Semi-Automated Curation Workflow
Year: 2024
Type: article
Abstract: Background: The environmental health vocabulary (EHV) is a controlled vocabulary developed by the US Environmental Protection Agency’s (EPA) Chemical Pollutant Assessment Division (CPAD) within the Office of Research and Development (ORD). The purpose of the EHV is to standardize the evidence supporting human health chemical assessments. The EHV is specific to the EPA’s CPAD and was initially developed by manual review and curation of terms extracted into the EPA Health Assessment Collaborative (HAWC). Standardization of endpoint terms improves the efficiency of data reporting, aggregation, filtering, and visualization. Recognizing that the current manual curation approach to updating the EHV does not scale with the rate new terms are entered into HAWC and the need to update existing HAWC terms with additional metadata, a semi-automated curation workflow was realized. Objectives: The objectives of this work are to describe both the manual creation of the current EHV, a new curation method for adding new terms to the EHV, and a method to update existing EHV terms with metadata and definitions using one semi-automated curation workflow. An additional objective is to minimize the requirement of manual review and analysis of extracted terms by applying a sequence of computational text analysis steps with a high level of accuracy. Methods: To create the EHV, terms extracted into HAWC were collected and manually curated into a hierarchical controlled vocabulary fit for use by EPA CPAD during assessment development. The new semi-automated workflow adds computational steps to add new terms, update existing terms, and add metadata. Specifically, the new EHV curation process includes application of a sequence of computational text analysis and manual steps as follows 1) a computational processing step to normalize extracted terms and identify matches to existing EHV endpoints, 2) manual quality assurance (QA) of the computationally identified matches using hierarchical data to help bin terms; 3) manual review of unmatched terms to identify possible matches not identified by computational processing; and 4) curation of EHV data, including the standardization of terms, completion of hierarchical data and metadata. Discussion: The EHV is a standardized vocabulary that makes consistent the reporting (and data extraction) of health effect, study and exposure design, experimental model, etc. information across studies and assessments. The EHV has been a great benefit to EPA CPAD’s assessment teams as it facilitates aggregation, visualization, and dissemination of the information supporting human health assessments. The curation workflow described here will expand the EHV by capturing more information and concepts needed to track provenance and it is anticipated that the EHV may promote data reporting standardization in the environmental health community and create more opportunities for sharing extracted content. The views expressed in this presentation are those of the author and do not necessarily reflect the views or policies of the U.S. EPA
Authors: Angrish, Michelle, Burns, Scott, Cleland, Joshua, Foster, Caroline, Kovach, Samuel, Markey, Kristan, Schulz, Brittany, Shapiro, Andy, Taylor, Michele, Woodall, George, Watford, Sean
Institution: Environmental Protection Agency
Language: English
Subfield: Molecular Biology
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W6911271749