Application Feature Improvements in Support of Human Health Assessments: Optimizations for Epidemiology Data Extraction
Toxicity values derived for human health assessments are relied upon for decision making to protect human health and the environment. To increase efficiency and optimize resources required to review and extract relevant information from literature, systematic review methods are employed with tools that improve user interfaces and interactions (UI/UX); standardize data exchange formats; and utilize artificial intelligence for (semi-)automation.
The Health Assessment Workspace Collaborative (HAWC) is a content management system for human health assessments. Data extraction features are available for both animal toxicology and epidemiology studies. Data extractions are integrated with visualization capabilities and can be produced with minimal data processing. We describe recent updates made by the HAWC team in coordination with EPA epidemiologists to update data extraction features including updates to UI/UX and adding more flexibility of the available forms to accommodate partial extractions.
Forms for data extraction include: study design, chemicals, exposures and exposure levels, adjustment factors, outcomes, and quantitative results. The flexibility provided with these updates enables the development and storage of information for evidence maps, toxicological reviews, visualizations, and interoperability with other tools. For example, information can be automatically extracted using machine learning methods in a platform like Dextr and exported into HAWC. Dextr is a tool that allows users to apply machine learning models to semi-automate data extraction from full text.
HAWC continues to be the major content repository for human health assessments. The updates to HAWC epidemiology data extraction improves the efficiency at which information can be extracted from studies and readily incorporated into assessments. The updates also facilitate integration of external tools and methods.