Abstract

Systematic reviews are used as the ‘gold standard’ to evaluate healthcare,education, and social policies. They are integral to the clinical decision making of healthcare professionals, and funding decisions made by governmental agencies.The rapid growth in primary research has not been matched by a growth in the efficiency of producing systematic reviews and consequently evidence-based decision making is struggling to remain feasible. This body of research aimed to develop and evaluate strategies towards the automation of systematic reviews, so that secondary health research can be produced more efficiently and cost effectively. Biomedical databases offer different products which vary in scale and content and researchers should be prepared to search several databases rather than relying on a single database. The title-only screening developed during this research was shown to be effective and demonstrated similar reliability to both predictive screening tools and human screening, and could be used with other automation tools to assist with screening. Progress with automation tools will be accelerated once technical barriers are overcome, and by pursuing proof of concept technologies into consumer ready products and thoroughly evaluating automation tools for reliability.

Year Manuscript Completed

2017

Keywords

Abstrackr; Algorithm; Automation; Biomedical database; Citation screening; Deduplication; Expediting Evidence Synthesis; Machine learning; PICO based title-only screening; Rapid Review; Scoping Search; Semi-automation; Systematic Review Assistant-Deduplication Module; SRA; Systematic Review.

Primary Language of Manuscript

EN

01Front.pdf (238 kB)

Share

COinS