Methodology

Dataset for an AI project to automate quality assessment of systematic reviews

Skills / interests: Risk of Bias assessment, PICO annotation, Data extraction

Methodological skills / interests: Bias

This project aims to build a dataset of 2,000 systematic reviews for the development of an artificial intelligence (AI) tool to assess the quality of systematic reviews. The tool when finished, will be open access and free to all users. It is called WISEST. The assessment of a minimum of 50 systematic reviews will grant you co-authorship in our publications. You will assess the systematic reviews with ROBIS and AMSTAR-2 tools.

 

Before appraising/assessing the reviews for the dataset, training on ROBIS and AMSTAR-2 will be provided, followed by piloting until the assessments of the reviews are consistent with the established decision rules. We have developed very specific instructions how how to assess the systematic reviews using these two tools. An expert methodologist will check your assessments and provide feedback, so that all assessments are standardised.

 

We will publish a methods study comparing the two tools, a dataset paper, and a final tool paper, which you will be a co-author (after completing 50 assessments).

 

Ideal applicant

Ideally you are familiar with screening studies against eligibility criteria, assessing risk of bias of trials (Cochrane risk of bias tool) and assessing the quality of reviews (using AMSTAR 2 or ROBIS), have done some data extraction, and have critical thinking and problem solving skills. The work will be remote and we will be doing 1 to 2 weeks of training prior to starting data extraction. It would be great if you have already authored a systematic review, but this is not necessary. Skills/Interests: Risk of Bias assessment, PICO annotation, Data extraction, Screening and selecting studies, ideally experience with ROBIS and/or AMSTAR 2 (although not necessary), and ideally familiar with network meta-analysis (although not necessary) Time requirement: Minimum 2 assessments per week, although the more the better. We will send out weekly meeting updates about our progress, and zoom meetings will be held every 3 weeks. Attendance at the meetings is not necessary.

Think you've got what it takes to get the job done for Carole?

Log in
to respond

Questions & comments

I’m interested, how can I apply?

Default profile Esra Mustafa - 3 months ago

I’m Interested how can I participate?

Default profile Alaa Mohamed - 1 month ago

Hello Carole, I am interested in helping you out. I have previous experience with systematic reviews and I am currently a science student at McGill in Montreal.

Default profile Negin Amini - 2 weeks ago

Hi Carole, I believe that the tool will be very useful to help those who are dedicated to Evidence-Based Medicine like me.