PRISMA 2020 Compliance

Transparent systematic review reporting with AI-assisted screening

What is PRISMA 2020?

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement provides an evidence-based minimum set of items for reporting in systematic reviews.

  • 27-item checklist for transparent reporting
  • Flow diagram for study selection process
  • Evidence-based reporting guidelines
  • International standard for systematic reviews

PRISMA 2020 Compliance

The I-category pipeline implements PRISMA 2020 guidelines throughout the entire systematic review pipeline, ensuring methodological rigor and transparent reporting.

1. Identification

Records identified from databases and registers

  • Records from databases (Semantic Scholar, OpenAlex, arXiv): n = X
  • Records from other sources (manual search, citations): n = X
  • Total records before deduplication: n = X
Automation

Automatic API-based retrieval with metadata logging

2. Screening

Duplicates removed, records screened for relevance

  • Duplicates removed (by DOI, arXiv ID, title similarity): n = X
  • Records screened using AI-assisted criteria: n = X
  • Records excluded with documented reasons: n = X
Automation

AI-PRISMA 6-Dimension screening with human verification

3. Eligibility

Full-text articles assessed for eligibility

  • Reports sought for retrieval: n = X
  • Reports not retrieved (no PDF available): n = X
  • Reports assessed for eligibility: n = X
  • Reports excluded with reasons: n = X
Automation

Automated PDF download with retry logic and error tracking

4. Inclusion

Studies included in final synthesis

  • Studies included in systematic review: n = X
  • Reports of included studies: n = X
  • Studies available for RAG query system: n = X
Automation

Vector database construction with full-text indexing

AI-PRISMA 6-Dimension Screening

The pipeline uses a 6-dimensional relevance scoring system for systematic screening, combining AI efficiency with human oversight.

1. Population Relevance

Does the study focus on the target population?

Scoring:Binary (Yes/No) or Likert scale (1-5)

2. Intervention/Exposure Match

Does the intervention or exposure align with inclusion criteria?

Scoring:Binary or Likert with keyword matching

3. Outcome Alignment

Are the measured outcomes relevant to research question?

Scoring:Binary or Likert with outcome validation

4. Study Design Fit

Does the study design meet methodological requirements?

Scoring:Categorical (RCT, quasi-experimental, observational, etc.)

5. Time Period Compliance

Is the publication date within the specified range?

Scoring:Binary (within/outside date range)

6. Language/Access Criteria

Is the study in acceptable language with available full text?

Scoring:Binary (accessible/not accessible)

Knowledge Repository Mode

50% relevance threshold (inclusive for broad coverage)

Systematic Review Mode

90% relevance threshold (strict for evidence synthesis)

PRISMA Flow Diagram Generation

The pipeline automatically generates PRISMA 2020-compliant flow diagrams from your screening data.

  • Automatic generation from `screening_results.csv`
  • Exportable formats: PNG, SVG, PDF
  • Customizable labels and colors
  • Includes exclusion reasons with counts
  • Version control with timestamps
Command
python scripts/07_generate_prisma.py --project projects/2025-01-15_MyProject

PRISMA 2020 Checklist Coverage

The pipeline addresses key PRISMA checklist items through automated workflows:

#5

Eligibility criteria

Defined in config.yaml, enforced in screening

#6

Information sources

Documented in database_strategy.md

#7

Search strategy

Boolean queries logged in query_log.json

#8

Selection process

AI-PRISMA 6-D screening with audit trail

#10

Risk of bias

Optional integration with Agent B2 (Evidence Quality Appraiser)

#16

Results of syntheses

RAG query system with citation tracking

#17

Flow diagram

Auto-generated PRISMA 2020 diagram