Semi-automated identification of biomedical literature

Brown University, Providence RI

Investigators

Linked publications, trials & patents

Abstract

Project Summary/Abstract Semi-automated identification of biomedical literature In this year-long project, we aim to develop a state-of-the-art system to assist the efficient identification of literature for systematic reviews and related evidence-synthesis products, which are key methodologies of Evidence-based Medicine (EBM) and of patient-centered comparative effectiveness research. The need to modernize the methods and processes of EBM is pressing, in part because of the rate with which new information is generated. Currently, approximately 100 new trials are published every day. This suggests that, over time, more systematic reviews and related products are being completed, and that they will probably have to examine larger evidence-bases and need more frequent updating. We propose to develop and evaluate approaches for semi-automating the formulation of search queries and the identification of citations that are eligible for systematic reviews. The goal is to increase the efficiency, transparency, and replicability of the process, while minimizing errors. We have two aims: (1) Develop a system for semi-automating the development of literature searches and the identification of relevant literature in systematic reviews. We envision that a human will start from natural-language descriptions of eligibility criteria and a small set of relevant documents and, over several iterations, will interact with the system to create and refine a search strategy and rank the corpus returned by the search in terms of relevance to the eligibility criteria. (2) Evaluate this approach prospectively, by comparing it to the standard approach for 2 systematic reviews, and retrospectively, by re-constructing the searches in up to five already completed systematic reviews. We will disseminate our findings through numerous channels including, but not limited to: publication, presentation at conferences, and through exploring interest in its wider adoption by AHRQs Evidence-based Practice Center Program, Cochrane Collaboration, and other groups conducting systematic reviews.