Aspirin for acute migraine
Randomized adults compared aspirin with placebo during one attack.
AI-Powered Systematic Review Screening Platform
Screen Thousands of Research AbstractsAI pre-screens every study with a confidence score and written rationale; you verify the borderline calls. Blind team screening, conflict resolution, and PRISMA-ready exports included.
Starts with AI screening. You can continue manually if you prefer.
Try the AI Screener
Paste your rules and one abstract. Get an instant first-pass decision before creating an account.
Include, exclude, or maybe
A short reason tied to your criteria
No upload, project, or sign-up
Built for real systematic review workflows
RISPubMed txtBibTeXEndNote XMLCSVORCID loginPRISMA exportBlind screeningConflict resolutionAudit trailConfidence scoresWritten rationale
RISPubMed txtBibTeXEndNote XMLCSVORCID loginPRISMA exportBlind screeningConflict resolutionAudit trailConfidence scoresWritten rationale
RISPubMed txtBibTeXEndNote XMLCSVORCID loginPRISMA exportBlind screeningConflict resolutionAudit trailConfidence scoresWritten rationale
RISPubMed txtBibTeXEndNote XMLCSVORCID loginPRISMA exportBlind screeningConflict resolutionAudit trailConfidence scoresWritten rationale
Integrations
Keep your review moving
Bring your references into screening, then take your included studies into data extraction.
Zotero
Collection import
Import a personal or group collection with a Zotero API key. Start a manual screening project with duplicate references combined.
One-time, read-only import. Your API key is not stored.
Import from Zotero
Evidence Table Builder
From screening to extraction
Finish screening your references, then continue in Evidence Table Builder to extract data from your study PDFs and build evidence tables.
Upload your full-text PDFs in Evidence Table Builder to get started.
Explore Evidence Table Builder (opens in a new tab)How screening works
Upload once. AI pre-screens with rationale. You verify the borderline calls. Export a PRISMA-ready audit trail.
01
Upload
RIS or PubMed txt
02
AI triage
Decision + confidence
03
Human review
Uncertainty queue
04
Export
PRISMA / Excel / RIS
Choose Your Approach
Manual, AI-Assisted, or Hybrid Screening?
Start with manual screening for full control, or let AI accelerate your process. You can always combine both approaches.
Manual
TraditionalFull reviewer control with blinded screening, consensus building, and audit-ready documentation.
- Complete human oversight
- Blinded screening by default
- Multi-reviewer consensus
- Time: ~40-90 sec per abstract on average
AI-Assisted
SmartLet AI handle initial triage while you focus on borderline cases and final decisions.
- 45x faster initial screening
- Rationale for every decision
- Human review of uncertainties
- Time: ~1-2 sec per abstract on average
Hybrid
RecommendedBest of both worlds: AI efficiency with human expertise where it matters most.
- AI handles obvious cases
- Human review of edge cases
- Audit trail for all decisions
- 5x faster than manual alone
Meet Your AI Screening Assistant
Think of AI as your virtual colleague: always ready to help, never replacing your expertise.
Intelligent Triage
Reviews abstracts against your inclusion criteria and provides detailed reasoning for each decision.
Virtual Colleague
Works alongside your team, handling routine decisions while flagging complex cases for human review.
Transparent Process
Every AI decision includes rationale, confidence scores, and relevant text excerpts for verification.
Speed & Consistency
Process hundreds of abstracts in minutes while maintaining consistent evaluation criteria across all studies.
Human Oversight
All borderline cases go to an uncertainty queue. You maintain full control and can override any decision.
Audit Ready
Complete decision trail with timestamps, reasoning, and reviewer confirmations for regulatory compliance.
Clinic chart review
Retrospective notes from a single clinic, with no comparison group.
Pilot feasibility study
A small pilot. Eligibility against the protocol is not clear yet.
Pain freedom at two hours
Double-blind trial in adults. Pain freedom was the primary outcome.
Narrative treatment review
A narrative review of options, not a study of participants.
Mixed-age crossover trial
Crossover design with adults and adolescents in one sample.
- 012Include
- 012Exclude
- 012Maybe
Important Considerations
- •AI screening is designed to assist, not replace, human judgment
- •All AI decisions should be reviewed, especially for systematic reviews requiring high sensitivity
- •The system is trained on published research but may not capture all domain-specific nuances
- •For tips and tricks on how to optimize your screening, check out our documentation
Robust Guardrails and Human Oversight
AI accelerates screening. Human expertise keeps the review defensible. Scroll through the safeguards that sit between triage and export.
Conservative Thresholds
The AI prefers human review over a silent miss. High-confidence calls can move quickly; everything else stays visible.
Decision policy
Prefer human review over silent miss
Auto-exclude threshold≥ 90%
Uncertainty band60–89%
Forced human review< 60%
Uncertainty Queue
Borderline cases — typically 10–20% of studies — are flagged automatically so your expertise lands where it matters.
Uncertainty queue
Borderline studies wait for you
PMID 39211877Maybe · 58%
Eligibility unclear from abstract alone
PMID 40122815Maybe · 63%
Typically 10–20% of a run
Human Override
Accept, reject, or rewrite any AI decision in one click. Notes and reviewer identity stay attached for the audit trail.
Human override
One click to change any AI call
AI suggested
Exclude · 91%
Reviewer decision
Include
Note: population matches inclusion criteria
Transparent Rationale
Every decision includes reasoning, relevant excerpts, and a confidence score — ready for protocol appendices and peer review.
Transparent rationale
Every call ships with evidence
Include94% confidence
“RCT in adults with matching primary outcome; intervention and comparator align with protocol.”
Excerpt + score + timestamp ready for audit
See the safeguards in action
Complete Transparency in Every Export
Download your screening results with full AI rationale, confidence scores, and audit trails. Perfect for systematic review protocols and peer review.
Sample Excel Export
Real screening results showing AI decisions, rationale, and human oversight
screening_results_final.xlsx
| Study ID | Title | Decision | AI Rationale | Confidence | Reviewer |
|---|---|---|---|---|---|
| PMID_12345 | Effectiveness of intervention X in population Y... | Include | RCT design, target population matches criteria... | 94% | AI + Human |
| PMID_67890 | Observational study of intervention Z effects... | Exclude | Wrong study design: observational, not RCT | 87% | AI Only |
| PMID_11111 | Mixed methods study examining intervention... | Include | Uncertain; flagged for human review | 62% | Human Override |
| PMID_24681 | Systematic review: Treatment outcomes in pediatric patients... | Include | Systematic review, relevant population and outcomes | 91% | AI Only |
| PMID_13579 | Case report: Rare adverse event following treatment... | Exclude | Case report; does not meet inclusion criteria for study design | 96% | AI Only |
| PMID_97531 | Pilot study of novel therapeutic approach... | Maybe | Small pilot study; requires human assessment | 45% | Pending Review |
| PMID_86420 | Multicenter trial comparing treatment protocols... | Include | Large multicenter trial, high methodological quality | 98% | AI Only |
| PMID_75319 | Animal model study of pharmacokinetics... | Exclude | Animal study; not human participants | 99% | AI Only |
Common Questions About AI-Assisted Screening
Get answers to the most common questions about reliability, security, and best practices.
Important Note on AI Reliability
AI performance varies by topic and eligibility criteria, so it should be used as an assistant rather than a replacement for human expertise. Review the preflight sample and uncertain cases, and validate a sample of decisions—especially for high-stakes systematic reviews intended for publication or clinical guidelines.
Research workflow articles worth indexing
Practical articles on screening, methodology, and RIS workflows that give Google clearer paths from the homepage to your strongest editorial pages.
systematic reviewscreening tools
Best screening tools for a systematic reviewScreening tools for a systematic review do the same narrow job in different shapes: they take a pile of titles and abstracts and help two people decide what sta...
07/10/202613 min read
ZoteroEndNote
How to import from Zotero, EndNote, or a RIS fileStudy Screener screens titles and abstracts. It does not replace Zotero, EndNote, or Mendeley. You bring references in once, screen them, then export RIS or CSV...
25/09/20265 min read
systematic reviewscreening
The Spreadsheet That Looked Like a Christmas TreeI was working with a research group recently, and they showed me their screening process. It was entirely set up in Excel. Every inclusion and exclusion decisio...
21/09/20265 min read
Ready to start screening?
Join researchers worldwide who are streamlining their systematic reviews. Start with our free forever plan and upgrade when you need AI superpowers.