Your first review, step by step
From an empty screen to a review that is ready to screen, in about ten minutes, and what comes after.
On this page
A systematic review starts with a question and ends with a paper. IAMB.AI keeps both ends in sight: the road across the top of every review shows the eleven stages, where you are and what comes next.
Before you start
Have these ready:
- Your review question, in one sentence.
- Two to five research questions that the review will answer.
- Eligibility criteria: what makes a study in or out (population, intervention, comparison, outcomes, study design).
- The exported results of each database search (for example from IEEE Xplore, ACM Digital Library, Scopus or Web of Science), with the search string you used and the date.
1. Create the review
Open Reviews and choose New review. Give it a title; you can change it later. You are now its owner.
2. Fill in Config
The Config tab holds the review's settings. The essentials:
- Question and research questions. They guide the AI's extraction form and the paper's introduction.
- Eligibility criteria, one per line. Each line gets a label: I1, I2… for inclusion, E1, E2… for exclusion. Screeners and the AI give a verdict on every criterion.
- Exclusion reasons for full-text screening (the PRISMA diagram counts them).
- Reviewers per record: two is the standard for dual screening.
- Blind mode: on, so that nobody sees the others' votes until screening is done.
Details: Config: the protocol of your review.
3. Invite your team
Add co-authors and reviewers in Config. Co-authors you invite can use the review for free on the Student plan. You can also add AI reviewers, which screen blind like a person. See Roles and AI reviewers.
4. Import your searches
Open Import and upload one file per database search: RIS, BibTeX, PubMed (.nbib), Web of Science, CSV or EndNote XML. Give each its source, date and search string. Together they become the search log your paper reports. See Importing search results.
5. Merge duplicates
Open Duplicates. Exact duplicates (same DOI, PMID or title) merge in one click; check the likely ones, such as a preprint and its published version. See Finding and merging duplicates.
6. Start screening
Open Title/abstract and start voting: I include, E exclude, M maybe. Your co-reviewer does the same, without seeing your votes. See Title and abstract screening.
The Overview tab always says what is next and why. When in doubt, go there.
What comes after
| Stage | What happens | Read |
|---|---|---|
| Conflicts | Records where votes differ are discussed and resolved | Conflicts and discussions |
| Full text | PDFs are found or uploaded, then assessed against every criterion | Full text |
| Extraction | A form collects the data of each included study | Data extraction |
| Related work | Prior surveys are compared with your review | Related work |
| Manuscript | Each section is written from your data and notes, then built to PDF | Writing the manuscript |
| PRISMA and Checklist | The flow diagram and the 27-item checklist, from the records | PRISMA |
| Export | Data, references and the LaTeX project | Exports |
Prefer to let two AI reviewers run every stage first? See The autopilot.