A machine learning classifier called Tagger has been developed to predict whether an article is an RCT, based on its title, abstract, and number of authors. In this project, we will test Tagger’s predictive accuracy on 12,000 articles taken from 843 Cochrane systematic reviews. The goal of our project is: (a) to assess Tagger performance; (b) to suggest possible improvements to Tagger’s development team (our collaborators).
• Tagger: http://arrowsmith.psych.uic.edu/cgi-bin/arrowsmith_uic/RCT_Tagger.cgi
• Tagger has already been evaluated in this previously published paper: Cohen AM, Smalheiser NR, McDonagh MS, Yu C, Adams CE, Davis JM, Yu PS. Automated confidence ranked classification of randomized controlled trial articles: an aid to evidence-based medicine. Journal of the American Medical Informatics Association. 2015 Feb 5;22(3):707-17. https://doi.org/10.1093/jamia/ocu025
• This work is part of our NIH-funded grant project, Text Mining Pipeline to Accelerate Systematic Reviews in Evidence-Based Medicine, which runs through 2020: http://ischool.illinois.edu/research/projects/text-mining-pipeline-accelerate-systematic-reviews-evidence-based-medicine
• Systematic review is a process for synthesizing literature: Hoang L, Schneider J. Opportunities for computer support for systematic reviewing-a gap analysis. In International Conference on Information 2018 Mar 25 (pp. 367-377). Springer, Cham. https://doi.org/10.1007/978-3-319-78105-1_40
• Cochrane produces systematic reviews of medical literature: http://www.cochrane.org