Back

Systematic Review Search Strategy Development Tools: A Practical Guide for Expert Searchers

Paynter, R. A.; Adam, G. P.; Hedden-Gross, A.; Twose, C.; Voisin, C.

2025-07-11 health informatics
10.1101/2025.07.10.25331296 medRxiv
Show abstract

ObjectivesTo provide the expert searcher community with reliable reviews on systematic review search strategy development tools (SSDT). We conducted a descriptive comparative analysis of the free SSDT listed on the Systematic Review Toolbox1-2 (SR Toolbox) website. Study Design and SettingOur search expert panel is composed of five biomedical librarians currently active in the US Agency for Healthcare Research and Quality (AHRQ) Effective Health Care Program Evidence-based Practice Centers (EPC). Two panel members reviewed each included SSDT that met inclusion criteria with any assessment disagreements resolved via consensus. We created a table of included SSDT, categorized them by search development step, rated their performance across our key domains and added narrative reviews outlining how to use them. ResultsOf the 102 total search tools listed on the SR Toolbox search software webpage (initial search September 2021 N=78; update search February 2023 N=24 new tools), we determined 21 were search strategy development tools and eligible for review. We found the preponderance of the 21 included SSDT are designed for the initial strategy development steps (i.e., scoping, keyword/phrase and subject term discovery), though there are tools in most steps meeting minimum performance criteria, with the notable exception of strategy logic evaluation. Our reviews include a graphical table covering key performance summary ratings and a narrative section providing more detailed information. ConclusionWe created a quick and intuitive SSDT guide for expert searchers that is organized by search strategy development phase, evaluates key features, and provides advice on how to use the tool effectively.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.