
Evidence Synthesis & Systematic Review Support for Medical Librarians
Accelerate systematic reviews with AI-powered evidence screening and data extraction
What You Can Do
Streamline your systematic reviews by using Claude to screen abstracts against inclusion criteria, extract structured data from full texts, and synthesize findings across studies. Reduce manual screening time by 60-80% while maintaining rigorous documentation of your review methodology. Claude becomes your collaborative partner throughout the entire evidence synthesis workflow, from protocol planning to final data synthesis.
Features
Screen hundreds of abstracts against PICOS criteria and inclusion/exclusion rules in minutes, with documented rationale for each decision
Extract and standardize data fields from research papers—study design, population, interventions, outcomes, results—in consistent table formats
Evaluate methodological rigor using Cochrane RoB 2, GRADE, or JBI frameworks with supporting evidence and bias domain ratings
Organize study findings thematically, identify patterns across diverse designs, and generate strength-of-evidence statements
Generate systematic review protocol checkpoints, flow diagrams, and registration-ready documentation for PROSPERO
Maintain transparent logs of all screening decisions and extraction rules, supporting reproducibility and methods sections
Work with RIS, BibTeX, and JSON formats to ingest reference lists and maintain study-level metadata throughout review
Example Output
Example 1: Abstract Screening Summary
| PubMed ID | Title | Decision | Rationale |
|---|---|---|---|
| 38214567 | Metformin and cardiovascular outcomes in type 2 diabetes | INCLUDE | RCT design, population ≥18y with T2D, intervention: metformin, outcome: CV events |
| 38156234 | Insulin glargine in geriatric patients | EXCLUDE | Population: geriatric (age cutoff specified ≥60), intervention: insulin not metformin |
| 38091456 | Effects of GLP-1 agonists on HbA1c | UNCLEAR | Intervention partially overlaps (combinotherapy with metformin). Needs full-text review. |
Example 2: Data Extraction Table
| Study ID | Design | N | Mean Age | Intervention | Duration | Primary Outcome | Effect Size | p-value | Funding |
|---|---|---|---|---|---|---|---|---|---|
| Adams2022 | RCT | 450 | 58.3y | Metformin 1g BID | 24 months | HbA1c reduction | 1.2% (95% CI: 0.9–1.5%) | <0.001 | NIH |
Example 3: Risk of Bias Summary
| Bias Domain | Risk Level | Evidence | Rating |
|---|---|---|---|
| Selection bias | Low | Randomization sequence generated by computer, allocation concealment used | ✓ |
| Performance bias | Some concerns | Single-blind design; patients aware of intervention | ⚠ |
| Detection bias | Low | Outcome assessors blinded to allocation | ✓ |
| Overall Risk | Low | Most domains low risk; performance bias acceptable | Included in meta-analysis |
What's Included
- Screening Workflow Templates: Pre-built decision trees for PICOS frameworks, Cochrane Consumers and Communication criteria, and custom inclusion rules
- Data Extraction Schemas: Customizable templates for RCTs, cohort studies, qualitative research, and diagnostic accuracy studies with field validation
- Critical Appraisal Tools: Integration with Cochrane Risk of Bias 2, GRADE assessments, JBI checklists, and CASP instruments
- Synthesis Prompts & Frameworks: Guided prompts for thematic synthesis, narrative reviews, vote-counting meta-analysis, and complex interventions
- Documentation & Reporting Templates: PRISMA 2020 checklist, PROSPERO registration scaffold, flow diagram data, and supplementary tables ready for publication
- Quality Control Workflows: Prompts for inter-rater reliability checks, consistency verification across screeners, and duplicate-study detection
Who It's For
- Medical Librarians
- Clinical Research Scientists
- Evidence Synthesis Coordinators
- Healthcare Policy & Guidelines Developers
- Graduate Students & Doctoral Researchers
Best For
- Accelerating initial abstract screening at scale
- Standardizing data extraction across diverse study types
- Conducting methodological quality assessments
- Organizing complex syntheses with multiple outcomes
- Generating PRISMA-compliant documentation






