AI supported reviews | Global Collaboration

AI supported systematic reviews

Topic leader: Bruno M. Coimbra

Bruno M. Coimbra, Department of Methodology & Statistics, Utrecht University, The Netherlands & Department of Psychiatry, Universidade Federal de São Paulo (UNIFESP), Brazil

Please contact Bruno M. Coimbra, if you would like to initiate new review projects.

The GCTS is eager to make research available to all around the world and thus supports freely available methods for doing systematic reviews as ASReview .

Planning embarking on a systematic review? If so, you're likely familiar with the common challenge faced by old-school reviewers: the lack of time to review everything thoroughly. ASReview is here to assist you in identifying eligible papers for your systematic review and streamlining the process.

What is ASReview

ASReview is a AI-aided open source software with a simple goal: assist researchers to find eligible papers for systematic reviews using state-of-the-art active learning techniques.

Have you searched for records for your systematic review and ended up with a file potentially containing thousands of them? With ASReview, you simply provide a few relevant and irrelevant records to our Electronic Searching Assistant (ELAS). By doing so, you're training ELAS's machine learning algorithm to rank the records in your file that you haven't reviewed yet from most relevant to least relevant based on their content

Projects

Occurrence and Prevalence of Latent Trajectories of Posttraumatic Stress Symptoms: A Systematic Review and Meta-analysis of Growth Mixture Modelling Studies.

Project members

Bruno Messina Coimbra, Mirjam van Zuiden, Laurens de Bruin, Rutger Neeleman, Beth Grandfield, Rens van de Schoot

Summary

Growth Mixture Modeling (GMM) identifies latent subgroups, or classes, representing similar patterns of symptom progression over time. Research has identified four main trajectories for posttraumatic stress symptoms (PTSS): chronic, delayed-onset/worsening, recovery, and resilience, with resilience often the most common response. Despite an increase in PTSS GMM studies, the occurrence and prevalence of these trajectories remain unexplored in meta-analyses. This study will systematically review and meta-analyze the occurrence and prevalence of latent PTSS trajectories in GMM studies. We will perform a systematic search using manual screening and active learning software ASReview in a large dataset ( > 12,000 records). Studies assessing PTSS at least three times post-trauma will be included. Random-effects meta-analyses will estimate the prevalence of PTSS latent class trajectories.

A systematic review and meta-analysis on the effectiveness of massed treatment in reducing PTSD symptoms

Project members

Bram Kemmere, Mayaris Zepeda Mendez, Bruno M. Coimbra, Miranda Olff, Mirjam Mink-Nijdam

Summary

Intensive treatments for PTSD might have similar clinical effectiveness but reduce symptoms faster and potentially decrease drop-out as compared to standard weekly treatment. The goal of the systematic review is to investigate how effective massed trauma-focused treatment is in reducing PTSD symptoms. In the meta-analysis, we aim to investigate how effective massed treatment is in reducing PTSD symptoms when compared to spaced treatment. Secondary goals are to investigate what variables potentially moderate treatment outcome. The inclusion criteria are: self-defined intensive treatments for PTSD (or synonyms of intensive treatment), adult patients, self-report and clinical interviews allowed. The exclusion criterium: not above the clinically significant threshold. We will search the following databases: MEDLINE / Pubmed, PsycInfo, SCOPUS, PTSDpubs, Web of science, Embase. Using ASReview, we will screen title and abstract. The default algorithm and settings of ASReview will be used and trained by the first authors. After that, full-text reading will be done using the Cochrane extraction sheet. Risk of bias assessment will be done with the Cochrane risk of bias tool. For the main outcome of the meta-analysis (PTSD symptoms), we will perform a random-effects meta-analysis with 95% confidence intervals (CI) to pool the standardized mean differences (SMD), using Hedges' g, to determine if there is a significant statistical difference between massed and spaced treatments pre- and posttest for PTSD symptoms.

Literature

ASReview LAB developers . 2022. ASReview LAB - a tool for AI-assisted systematic reviews (v1.1) Zenodo. - DOI
Rens van de Schoot, Marit Sijbrandij, Sonja D. Winter, Sarah Depaoli & Jeroen K. Vermunt (2017) The GRoLTS-Checklist: Guidelines for Reporting on Latent Trajectory Studies, Structural Equation Modeling: A Multidisciplinary Journal, 24:3, 451-467, DOI: 10.1080/10705511.2016.1247646
Rens van de Schoot, Marit Sijbrandij, Sarah Depaoli, Sonja D. Winter, Miranda Olff & Nancy E. van Loey (2018) Bayesian PTSD-Trajectory Analysis with Informed Priors Based on a Systematic Literature Search and Expert Elicitation, Multivariate Behavioral Research, 53:2, 267-291, DOI: 10.1080/00273171.2017.1412293

How?

Check out the video.

For help and questions reach out to asreview@uu.nl or go to our Github page on https://github.com/asreview

For more information and to download the ASReview, please visit https://asreview.nl/

ASReview is an open-source project coordinated by Rens van de Schoot (full Professor at the Department of Methodology & Statistics and ambassador of the focus area Applied Data Science at Utrecht University, The Netherlands), together with Jonathan de Bruin (Lead engineer of the ASReview project and working at the Information and Technology Services department at Utrecht University).

ASReview workshop in Cape Town

At the 1st Global Collaboration on Traumatic Stress Conference preconference workshop , “Using AI to quickly get a systematic overview of the literature” was held on the afternoon of 12 December at Stellenbosch University Faculty of Medicine and Health Sciences. The workshop was presented by three members of the ASReview team, Rens van de Schoot, Duco Veen and Jelle Teijema of Utrecht University. The workshop provided an overview of ASReview, which uses state-of-the-art active machine learning techniques to solve the challenge of screening large amounts of text, and guided participants through hands-on exercises covering set up, labelling of articles for inclusion and exclusion, and interpretation of the review results. The workshop proved extremely popular with the original limit of 40 participants waived to accommodate the enthusiastic interest. The approximately 55 participants included local and international GCTS participants, as well as researchers and students from Stellenbosch University, the South African Medical Research Council, and other local institutions.