SciSpace Academy
Tutorial

How to Conduct a PRISMA-Ready Systematic Literature Review with AI

Avi StaimanEnglishSystematic review

TL;DR

1-min read

A rigorous systematic literature review can take weeks of searching, screening, data extraction and reporting before a PRISMA-compliant manuscript is ready. In this tutorial, Avi Staiman, Founder and CEO of Academic Language Experts, member of the UN AI for Developing Countries Forum (AIFOD), core member of CANGARU representing EASE and an editor at The Scholarly Kitchen, presents a practical, guided method for getting from a research question to structured, PRISMA-ready outputs more quickly. It is aimed at researchers, graduate students and evidence-synthesis teams who need a repeatable process that will stand up to peer review.

Key takeaways

  • A well-framed research question is what drives a structured, PRISMA-aligned review from the outset.
  • PRISMA sets out what a review should document, including the search strategy, screening decisions and a flow diagram of records identified, excluded and included, so keeping these records from the start makes reporting easier.
  • AI can speed up labour-intensive stages such as surfacing candidate papers, screening and summarising content.
  • Insights from included studies should be extracted in an organised, traceable way that supports the final report.
  • Integrity is maintained when researchers stay responsible for eligibility criteria, screening decisions and data verification, and disclose how AI was used.
  • Structured workflows and AI help shorten searching, screening and extraction, but planning, protocol writing and quality checks still need dedicated researcher time.
Apply it

To apply this, set up your next review with a clear question and a record of every screening decision, then use SciSpace for searching and extraction while you verify each step.

How to do it

  1. Write the review protocol and define eligibility, outcomes, sources, dates, and synthesis.

    Write the protocol and define the question, eligibility criteria, outcomes, sources, dates, screening process, appraisal method, and synthesis before retrieving records.

  2. Review and save every database search strategy before execution.

    Save the complete search strategy for each database or source, including fields, Boolean logic, filters, date run, and number retrieved. These records support PRISMA-S reporting.

  3. Deduplicate and screen records while retaining decisions and exclusion reasons.

    Deduplicate and screen records while preserving reviewer decisions, conflicts, full-text exclusions, and standardized reasons. Keep counts synchronized across stages.

  4. Verify extractions and quality assessments against full texts.

    Verify extraction and risk-of-bias judgments against full texts and retain the evidence location. Pilot the forms before applying them to every included study.

  5. Generate the PRISMA flow, methods details, tables, and checklist from the documented workflow.

    Generate the flow diagram, methods details, study table, appraisal summary, and checklist from the documented record. Resolve count discrepancies before publication.

Watch the tutorial41 min

How to Conduct a PRISMA-Ready Systematic Literature Review with AIPlay video: How to Conduct a PRISMA-Ready Systematic Literature Review with AI

Chapters

  1. How to start a systematic review in SciSpace Agent (PRISMA mode)
  2. How an AI agent runs a PRISMA systematic review: the 12-step plan
  3. How to generate PICO research questions with AI
  4. How to set inclusion and exclusion criteria with AI
  5. Systematic review search strategy: PubMed, Google Scholar and Zotero
  6. Do AI literature review tools read the full text? Open-access bias
  7. How AI removes duplicates and re-ranks search results
  8. Reviewing search results and relevance scores
  9. What is title and abstract screening? AI screening demo
  10. How full-text screening works in an AI systematic review
  11. How to extract data from papers for a systematic review with AI
  12. How to generate a PRISMA flow diagram and report with AI
  13. Can you submit an AI-generated systematic review without checking it?
  14. SciSpace Agent skills: grants, slides, websites and more
  15. How to search clinical trials with AI: EU lung cancer example
  16. How to tell if you are using AI responsibly in research

Try this on SciSpace

Frequently asked questions

What does PRISMA-ready mean for a systematic review?

PRISMA is a reporting guideline that sets out what a systematic review should document, including the search strategy, screening decisions and a flow diagram showing how many records were identified, excluded and included. A PRISMA-ready review keeps these records from the start, so the final manuscript can report every step transparently and meet journal expectations.

Can AI be used in a systematic review without harming integrity?

Yes, if researchers stay responsible for eligibility criteria, screening decisions and data verification, and report how AI was used. AI is best applied to labour-intensive tasks such as surfacing candidate papers or summarising content, with human checks on every decision that affects the evidence base. Transparent disclosure lets reviewers assess the approach.

How long does a systematic literature review usually take?

Traditional systematic reviews often take many weeks or months, depending on the scope, number of records and team size. Searching, screening and extraction typically take the most effort. Structured workflows and AI assistance can shorten these stages, but planning, protocol writing and quality checks still require dedicated researcher time.

Related tutorials