Systematic Reviews in Elicit

With AI and language models, Elicit can help you save up to 80% of the time on a systematic review without sacrificing accuracy. The Systematic Reviews workflow provides step-by-step guidance through search, screening, and data extraction, culminating in a research report summarizing the relevant papers.

How It Works

Elicit supports the major steps of a systematic review:

1. Set up your review on the Protocol Page
2. Gather all papers for the systematic review
3. Screen papers by title and abstract
4. (Optionally) Screen papers by full text
5. Extract data from screened-in papers
6. Generate a research report

Set Up Your Review: The Protocol Page

When you start a new systematic review, the first thing you'll see is the Protocol Page. This page shows every step and setting in one place, so you can configure your review before anything runs.

Research question

Enter your research question at the top. Elicit uses this to generate search strategies, screening criteria suggestions, extraction column suggestions, and to guide the research report. You can update it at any time.

Additional Context (for protocol details)

Use this free-text field to describe your eligibility criteria, PICO framework, or other protocol parameters. For example:

• Population, Intervention, Comparator, Outcome definitions
• Inclusion and exclusion criteria
• Study design requirements (e.g., "RCTs only" or "exclude case reports")

This gives Elicit additional context to apply throughout your review.

Workflow stages

The Protocol Page displays every stage of your review as a visual timeline. Some stages are always included, and others can be toggled on or off:

Always on:

• Gather — search for and collect papers
• Extraction — extract data from included papers
• Report — generate a research report

Toggleable:

• Abstract screening — screen papers by title and abstract. Turn this off if you want all gathered papers to go directly to extraction.
• Full-text screening — an additional screening step using the full paper text. Only available when abstract screening is enabled. Useful when your eligibility criteria require details not found in abstracts.

Search strategy

Choose how Elicit gathers papers:

• Semantic search — Elicit interprets your question and finds conceptually relevant papers. Best for exploratory reviews.
• Keyword search — use Boolean strings (AND, OR, NOT) for precise, reproducible searches. Best for PRISMA-compliant reviews or HTA filings.

Note: you can add additional search options once on the Paper Sources step.

Additional settings

• Strict screening — when enabled, papers that fail any strict criterion are automatically excluded. Useful for hard requirements like study type or date range.
• Figure extraction — allow Elicit to analyze figures and images in papers during screening and extraction. Available on Scale and Enterprise plans.
• Report template — select the structure and format of your final research report.
• Dual review — invite another reviewer to independently make screening decisions in Elicit, on abstract and full-text screening. Available for Enterprise plans. To enable, toggle it on below the "Additional context" field and invite a second reviewer.

Modifying your protocol mid-review

You are not locked into your initial settings. Click Protocol in the sidebar at any point to reopen the Protocol Page and make changes.

• Changes apply to stages that haven't completed yet.
• If you enable a stage that runs before your current stage (e.g., turning on full-text screening while you're in extraction), you will be taken back to that earlier stage.
• If you disable a stage that has already run, its effects are removed from downstream stages. The results are saved, so this is reversible — re-enable the stage and they come back.

Gather Relevant Papers

The Gather step consolidates papers into a single systematic review. You choose an initial search strategy on the Protocol Page, but you can add more searches, delete tabs, or change your approach within the Gather stage itself. Per systematic review Pro users can add up to 5,000 papers, Scale users up to 20,000 papers, and Enterprise users up to 40,000 papers. Additionally, the final report generated at the end of the systematic review workflow includes the 80 papers with the highest screening scores by default. However, this limit can be extended based on the subscription plan: Pro Plan allows up to 135 papers, and Scale Plan allows up to 200 papers in the final report.

• Multi-Tab Search: Open multiple search tabs using the + icon. Combine broad semantic questions with precise Boolean queries simultaneously.
• Iterate & Refine: Test different search strategy variations. Elicit automatically de-duplicates papers across active search tabs.
• Uploads: Upload PDFs directly or select papers from your Elicit Library.
• Multiple databases: Search across Elicit's internal corpus of 138M papers, PubMed, and ClinicalTrials.gov.
• Removing Automatically Added Papers: If Elicit adds a search tab with papers it finds relevant, you can remove it by clicking the X on that tab. This ensures only your chosen sources remain.
• Limiting Reviews to Uploaded Papers: To use only your uploaded papers, delete any automatically added search tabs and ensure only the tab with your uploaded papers remains active.

Once you've gathered all relevant papers, click "Define screening" to continue. Before proceeding, ensure that only the desired paper sources remain active by reviewing and managing the search tabs as needed.

Screening

The screening phase refines your criteria using a sample of papers. Title and abstract screening runs first, followed by full-text screening if you enabled it on the Protocol Page.

Defining criteria:

• Elicit auto-generates screening criteria based on your research question and protocol details
• You can edit, disable, or manually add criteria
• Mark individual criteria as strict to automatically exclude papers that fail them

Evaluating results:

• Click Evaluate screening when you're satisfied with your criteria
• Papers failing strict criteria appear at the bottom of results
• Other papers sort by likelihood of meeting all criteria, with screening scores visible
• Click any paper to see its per-criterion decisions, exclusion reasons, and source quotes from the abstract supporting each decision
• To override a screening decision on a specific paper, click on that paper and then select your decision (Exclude or Include) and then press Save & next. If you exclude a paper, you can choose an exclusion reason as well.

Threshold adjustment:

Once screening has completed, increase or decrease the screening threshold to include or exclude more papers. Higher thresholds are more selective. When you've finished screening, click "Define extraction" to continue.

Note: if all of your screening criteria are set to Strict, the screening threshold will not be displayed as it's not relevant to strict criteria.

Data Extraction

You'll first define extraction columns on a subset of papers to make sure your columns are functioning as you desire. Elicit generates suggested data extraction columns based on your research question. You can edit these suggestions, remove them, or add your own columns.

Click "Run extraction" when you're satisfied with your column definitions.

Click on any cell in the finished data extraction table to view supporting quotes from the paper so that you can easily check the AI-generated answers for accuracy.

For Scale and Enterprise users, extractions can pull from figures, tables, charts, and diagrams in addition to text.

Research Report

The final step generates a research report summarizing your review. When more than 80 papers are screened in, the 80 papers with the highest screening scores are included in the report. This default limit can be increased with higher-tier plans: Pro Plan allows up to 135 papers, and Scale Plan allows up to 200 papers in the final report.

If you selected a report template on the Protocol Page, the report will follow that structure.

Other Key Features

• Browser Extension: Use your institutional database access to pull full texts before data extraction.
• CSV Export: Export results from any step for further cleaning or analysis.
• Collaboration: Team and Enterprise plans enable real-time collaborative editing. View-only shareable links are available on all plans.

Systematic Reviews are also sometimes called Systematic Literature Reviews, SRs, or SLRs. They are available on Pro and higher plans.