Using Elicit for a Systematic Reviews or Literature Reviews

Elicit can help with each stage of paper collection and analysis:

  1. Find papers

  2. Screen papers for inclusion

  3. Systematically extract data from papers

Go through all of these steps rigorously for a high-quality meta-analysis, or just quickly find some papers to bolster the related work section of your paper. Utilize both papers Elicit finds for you and papers you may have already gathered that you can upload to your Elicit library.

Subscribe to our YouTube channel to make sure you don't miss any new video tutorials or updates.

How Elicit can help with each stage of a review:

Find papers

You can run multiple Find papers searches to find relevant papers on a topic. For each search, continue clicking "Load more" until you cease to find relevant papers. See here for more tips on Find papers.

Once you find papers, download the PDFs (either on Elicit or via your institutional access), then upload them to your Elicit library for use in the later stages.

Screen papers for inclusion

Extract data from PDFs using the paper PDFs in your Elicit library. For each of your inclusion criteria, add a column to Elicit to check whether each paper meets the criterion. Most likely you will need to use a custom column tailored to your review. Some tips:

  1. Most important: Experiment with how you define the screening columns. Especially if you're screening many papers, it's worth figuring out what works best

  2. Try multiple different ways of defining the column:

    1. ask your yes/no screening question, e.g. "Was the study performed in Asia?"

    2. ask Elicit to extract the relevant info, e.g. "Where was the study performed?"

  3. When you define the screening columns, ask Elicit to only say that a paper should not be included if it's very certain. That way you can avoid missing relevant papers. You can follow up with some manual review to clear out any papers you don't want. (To make manual review easier, start with the quotes provided by Elicit.)

  4. Don't use high-accuracy mode unless you really need it, to save on cost

Systematically extract data from papers

The same advice applies as for "Screen papers for inclusion", except that you should likely use high-accuracy mode. For data extraction, the gain in accuracy is usually worth the cost.