Objectives: The unprecedented flood of items relating to Covid-19 makes it crucial for users to find effective ways to filter their searches. To this end, several different groups have released search filters designed to focus a user’s search on topics such as diagnosis, critical care or “long covid” syndrome. The objective of this project is to examine how effective these filters are in terms of sensitivity and selectivity.
Methods: This project will proceed in stages, beginning with the selection of publicly available Covid-19 search filters. Once those are selected, team members will collaborate on a set of rubrics to be used for testing the concept behind each filter (i.e. what is meant by “transmission” in this context, etc.).
Once grouped sets of filters and rubrics are defined, a very broad search strategy will be constructed for each set. These searches will be performed against PubMed, and a random selection of resulting items will be retrieved for examination by team members. For each randomly selected set of results, members will examine each item and determine whether it matches the predetermined rubric. Those matching items will comprise “known good” validation sets that can later be used to test each search filter in a quantitative manner.