B1075
Title: In the pursuit of automating meta-analysis
Authors: Eleni Elia - Oxford Brookes University (United Kingdom) [presenting]
Abstract: Published research examining the same or similar research questions often reaches conflicting findings. Findings can be biased for various reasons including, but not limited to, small effect size and small sample size. Meta-analysis studies can be better powered to provide evidence and an overall answer that is of low bias. Typically, meta-analysts are handed over the relevant estimates to conduct a meta-analysis. However, these estimates need to be manually extracted from the identified studies to be included in the meta-analysis. However, in the era of automation and rapid advancements, adoption of generative AI, and natural language processing, the process of extracting the relevant meta-analysis estimates from publications can be automated to not only speed up the process but also to contribute to the need for providing up-to-date evidence, to provide an informed, updated effect size estimate related to the question of interest. Leveraging these advancements is being explored to expedite the synthesis of published research findings to offer a timely, informed evidence base.