Identifying Relevant Evidence for Systematic Reviews and Review Updates

Systematic reviews identify, assess and synthesise the evidence available to answer complex research questions. They are essential in healthcare, where the volume of evidence in scientific research publications is vast and cannot feasibly be identified or analysed by individual clinicians or decision makers. However, the process of creating a systematic review is time consuming and expensive. The pace of scientific publication in medicine and related fields also means that evidence bases are continually changing and review conclusions can quickly become out of date. Therefore, developing methods to support the creating and updating of reviews is essential to reduce the workload required and thereby ensure that reviews remain up to date.
This research aims to support systematic reviews, thus improving healthcare through natural language processing and information retrieval techniques. More specifically, this thesis aims to support the process of identifying relevant evidence for systematic reviews and review updates to reduce the workload required from researchers.
This research proposes methods to improve studies ranking for systematic reviews. In addition, this thesis describes a dataset of systematic review updates in the field of medicine created using 25 Cochrane reviews. Moreover, this thesis develops an algorithm to automatically refine the Boolean query to improve the identification of relevant studies for review updates.
The research demonstrates that automating the process of identifying relevant evidence can reduce the workload of conducting and updating systematic reviews.