SOLUTIONS > The National Institute of Environmental Health Sciences
The Division of Translational Toxicology (DTT) is a part of The National Institutes of Health, the primary agency of the United States government responsible for biomedical and public health research. DTT scientists use a variety of traditional and cutting-edge approaches to better understand how factors in the environment may impact humans’ health, including literature reviews.
Conducting literature reviews can be time-consuming and labor-intensive. The data extraction stage is especially challenging in environmental health science, due to a multitude of evidence streams and complex study designs.
DTT needed a flexible, web-based tool for semi-automated data extraction. Code named project Dextr, the collaboration between DTT and Evidence Prime yielded a solution based on the Laser AI technology and optimized for environmental health research.
The team then put the tool to test in a prospective study closely resembling typical DTT workflow. Time savings and quality (sensitivity and precision) were measured in both manual and semi-automated modes. The results show 53% reduction in extraction time while maintaining similar recall and precision rates. Based on these results, DTT decided to implement the solution in their evidence mapping projects.
The solution based on Laser AI provides similar performance to manual extraction in terms of recall and precision and greatly reduces data extraction time. Unlike other tools, it provides the ability to extract complex concepts (e.g., multiple experiments with various exposures and doses within a single study), properly connect the extracted elements within a study, and effectively limit the work required by researchers to generate machine-readable, annotated exports. Furthermore, the solution addresses data-extraction challenges associated with environmental health sciences literature with a simple user interface, incorporates the key capabilities of user verification and entity connecting, provides a platform for further automation developments, and has the potential to improve data extraction for literature reviews in this and other fields.
Furthermore, the solution:
