We used interviews in order to investigate the technologies reviewers use and why, based on our interviewees’ detailed explanations. We conducted a series of semi-structured interviews with 16 systematic reviewers who had co-authored at least one published systematic review. Yet to understand whether and how reviewers are actually using these tools, and how well the current and emerging technologies fulfill reviewers’ requirements, a gap analysis is needed. A 2014 review listed fourteen tasks that could potentially be automated, and identified more than 10 applications being developed to assist different phases of the review process, including search engines (Quick Clinical and Metta) and data extraction support tools using machine learning and natural language processing (ExaCT and RobotReviewer). Currently, research prototype systems are in development to support or automate each of the steps shown in Figure 1. The Systematic Review Toolbox 6 collects and describes relevant tools. Another end-to-end tool, EPPI-Reviewer 5, provides (and continues to develop) advanced features such as automatic term reorganization, and document clustering and classification, using machine learning and data mining. Some commercial products are designed as end-to-end support tools: DistillerSR 3 and Covidence 4 primarily provide an integrated environment for data capture and management, for tasks such as harvesting search results from databases, screening studies, and providing questionnaires for manual data extraction. Previous survey research has found that reviewers typically use software such as EndNote, Reference Manager, RefWorks, and Excel to manage references. This intense cost in time and effort has led to the development of computer support tools. Steps in a systematic review process according to What current computer support technology are reviewers using? What technical problems and challenges do reviewers face in conducting a systematic review? Through interviews with systematic reviewers, we seek to identify the gaps between the computer support available and what reviewers actually use, at a Research I university without an academic medical center. However, the gap between reviewers’ current practices and existing computer support is not well understood. Dissemination of tools and methods is an ongoing effort (for instance by the Medical Library Association and by Cochrane ) and there are some large-scale efforts to transform the production of systematic reviews (e.g. Already, several commercial software packages have been designed as end-to-end tools to support the reviewing process. To address this, informatics and methodology researchers are working to minimize the effort required to complete systematic reviews. Despite their importance, systematic reviews require great amount of human effort: a mean of 67 weeks from deposit of a protocol to publication of the review, with a mean of 1000 hours of person time. Systematic reviews can support translation of research into practice, when the underlying research has concordant findings and they can also draw attention to gaps in the evidence, such as discordant findings that need further investigation. A systematic review is a type of literature review designed to provide all available evidence on a given question.
0 Comments
Leave a Reply. |