Meeting Register Page

ESMARConf2023: Workshop 4 - Testing (semi)-automated de-duplication methods in evidence synthesis

Organiser: Kirsten Ziesemer

When searching for a literature review it is required to search in multiple bibliographic databases with overlapping content. Removal of these duplicate references is essential to reduce reviewer workload when screening for relevant abstracts. Moreover, proper removal of duplicate references avoids the unintended removal of eligible studies, limiting potential bias to the literature review. Duplicate removal or de-duplication is a time and resource constraining process of evidence synthesis which (semi) automation of this task could reduce. The purpose of this interactive workshop is a run through of de-duplication methods using R and discuss and reflect on best practices during (semi) automated duplicate removal. Based on a systematic literature search and a national de-duplication workshop, we identified several (semi) automated de-duplication methods. In this interactive workshop we will perform a de-duplication on an available small dataset (e.g. 1000 references from three databases) using R. We will compare methods based on performance using a benchmark dataset (i.e. compare number of true positives, false positives, true negatives, false negatives, precision and sensitivity) and results from the national de-duplication workshop. Ideally, strategies for improving de-duplication procedures using R in evidence synthesis will be formulated during this interactive workshop.

Learning objectives:
- Overview of performance of (semi) automated de-duplication methods
- Run-through of R-based de-duplication method (ASySD and/or revtools)
- Discuss and reflect on best practices during (semi)-automated duplicate removal
- Formulate strategies for improving de-duplication procedures in evidence synthesis

Target audience:
No prior experience needed 

Prerequisites:
Rstudio installed

Mar 28, 2023 11:00 AM in London