Systematic review screening tool

Screen 5,000 papers in an evening.

Your search returned hundreds or thousands of citations. Lumina turns that pile into a title and abstract screening queue, ranks records against your criteria, and keeps Include, Exclude, Maybe, notes, reviewer conflicts, and exports together while you screen.

For the post-search screening slog
Do not read abstracts in random export order
No separate spreadsheet cleanup later
Live screening queue AI-assisted ranking
Lumina screening interface showing AI relevance scores, abstract details, and reviewer decisions
1

Define criteria

Put the inclusion and exclusion rules next to the records reviewers are about to judge.

2

Find & import

Move PubMed, Scopus, Web of Science, RIS, or CSV records into the same screening queue.

3

Screen abstracts

Review the abstracts most likely to matter first, then continue through the rest of the queue.

4

Resolve & export

Leave screening with decisions, notes, conflicts, and PRISMA counts already structured.

How Lumina works

What happens after your database search returns 5,000 records?

That is the moment Lumina is built for: import the records, prioritize likely matches, screen titles and abstracts, resolve disagreements, and export the decisions you need for reporting.

Lumina database import builder for searching scientific sources
Database search & import

You finished the search. Now get the citations into one place.

Instead of splitting records across downloads and spreadsheets, Lumina gives the review a single queue that is ready for title and abstract screening.

  • Bring in records from PubMed, OpenAlex, Semantic Scholar, CrossRef, RIS, or CSV.
  • Preview titles before import so obvious misses do not become reviewer work.
  • Keep each imported citation attached to the review project it belongs to.
Prioritized title and abstract screening

The painful part is reading thousands of abstracts in no meaningful order.

Lumina ranks records against your criteria, so the first screening session starts with the abstracts most likely to matter instead of whatever row happened to come first in the export.

  • Sort by likely relevance so reviewers do not burn energy on obviously weak matches first.
  • Use Include, Exclude, Maybe, notes, and keyboard shortcuts in the same screening view.
  • Open Pixel-Bot when a reviewer wants to know why an abstract looks relevant.
Lumina screening tool with relevance score and reviewer decision controls
Lumina database import results list with selected studies ready for import
Decision trail

Do not let decisions, notes, and PRISMA counts become a second project.

The reporting mess starts when decisions live in one sheet, notes in another, and PRISMA counts somewhere else. Lumina keeps them connected while reviewers screen.

  • See decided, unresolved, excluded, and waiting-on-reviewer records in the project.
  • Handle two-reviewer disagreements before they disappear into email threads.
  • Export screening decisions and PRISMA flow data when it is time to write.
Problems researchers recognize

If your screening work sounds like this, Lumina fits.

Lumina is not a generic research workspace. It is for the title and abstract screening stage where record volume, reviewer fatigue, AI trust, conflicts, and exports collide.

Import

“My search exported hundreds of records. Now they are split across files.”

Bring database searches, RIS exports, and CSV files into one review project before screening starts.

Database import builder crop

Prioritize

“We are screening in spreadsheet order, not relevance order.”

Rank titles and abstracts against your criteria so likely includes reach reviewers earlier.

AI prioritized screening crop
AI

“I do not want AI deciding inclusion for me.”

Lumina ranks and explains. Reviewers still make the Include, Exclude, or Maybe decision.

2x

“Two reviewers disagree, and we find out too late.”

Run independent screening, compare decisions, and resolve conflicts inside the review workflow.

PR

“Writing the methods means rebuilding what happened.”

Keep decisions, notes, and PRISMA counts structured while you screen, then export them when you write.

Reviewer-controlled AI

Use AI for triage, not as a replacement reviewer.

Researchers are right to be cautious about AI in systematic reviews. Lumina uses AI to order the queue and explain possible matches, while the review team keeps control over inclusion decisions.

No black-box inclusion

AI can rank and explain records, but reviewers still choose Include, Exclude, or Maybe.

No orphaned notes

Keep decisions, notes, progress totals, and exports connected to the same project record.

Built for screening

Criteria, queues, imports, reviewer decisions, conflicts, and PRISMA counts stay in one workflow.

Know where review data lives

Lumina is hosted in the EU, so teams are not guessing where review records are stored.

Free Tool

Already have the counts and just need the PRISMA diagram?

Use the free tool when you only need to turn identification, screening, and inclusion counts into a PRISMA 2020 flow diagram. No signup required.

Create Diagram Free

Frequently Asked Questions

My database search returned thousands of records. Where do I start?

Start with the demo queue. It shows how Lumina ranks titles and abstracts, records Include/Exclude/Maybe decisions, and keeps notes attached before you upload your own records.

Will AI make inclusion decisions?

No. Lumina ranks records against your criteria and explains possible matches. Reviewers still make every Include, Exclude, or Maybe decision.

Can I bring in the searches I already ran?

Yes. You can search supported databases directly or upload RIS/CSV exports from sources such as PubMed, Scopus, Web of Science, Zotero, or EndNote.

Will I have to clean everything up in Excel after screening?

Yes. Screening decisions, notes, conflicts, and PRISMA counts stay structured during the review so they are ready to export when you write up results.

Try the moment after the search: 5,000 records, one screening queue.

Try the queue, AI relevance scores, Pixel-Bot explanations, and decision trail before you upload a real database export.