Empiric Testing and enhancement of web-based abstract screening tool(Abstrackr)

Brown University, Providence RI

Investigators

Abstract

EMPIRICAL TESTING AND ENHANCEMENT OF WEB-BASED ABSTRACT SCREENING TOOL (ABSTRACKR) In this year-long project, we aim to empirically assess the performance and efficiency of state-of-the-art information analysis technologies to assist the production of systematic reviews and meta-analyses that are increasingly being used as a foundation for evidence-based medicine and stakeholder-driven comparative effectiveness reviews. We have developed AbstrackrTM (hereon, Abstrackr), a human-guided computerized abstract screening tool that aims to reduce the need to perform a tedious but crucial step of manually screening many thousands of abstracts generated by literature searches in order to retrieve a small fraction potentially relevant for further analysis. Abstrackr makes use of machine learning techniques, and is offered as a free web-based tool that enables management of the screening process. We also aim to revise the web-interface of Abstrackr to make it more intuitive, user friendly, and add documentation and functionalities requested by users; and to revise Abstrackr?s back-end, which includes the way the software parses and analyses citations, fits machine learning models, and makes computations, to make it more efficient. These revisions will ensure that the tool becomes more robust, and that it remains usable for larger projects and for many teams. The proposed work will be carried out by the developers of Abstrackr, comprising a highly experienced team of systematic review investigators and computer scientists at Brown University and the University of Texas at Austin, who have been working together for at least seven years. We will pursue dissemination of the findings of this assessment and of the revised tool through numerous channels including, but not limited to publication, presentation at conferences, exploring interest in its wider adoption by the Agency for Healthcare Research and Quality Evidence-based Practice Center Program, Cochrane Collaboration, and other groups conducting systematic reviews. We will also continue to make all code available online. Our aims are to: Aim 1. Empirically measure the efficiency and accuracy of the prediction algorithms in Abstrackr in the computer-assisted semi-automated screening of citations for eligibility in systematic reviews. Aim 2. Improve and add to the functionality of the Web-based Abstrackr software, based in part on enhancements suggested by a panel of identified heavy users.