Workflow Overview

Identification & Screening

These first five steps take you from a research question to a finalized set of relevant papers.

Step	Name	What happens
1	Search Strategy	AI decomposes your question into PICO terms and generates database-specific queries
2	Import Files	Search PubMed directly or upload RIS/BibTeX/CSV from any database
3	Deduplication	4-layer duplicate detection: DOI, title, fuzzy, and semantic
4	TiAb Screening	AI screens every title & abstract with confidence + PICO evaluation
5	Human Review	Reviewers verify AI decisions, resolve conflicts

Analysis & Export

Once your paper set is finalized, these steps handle deep analysis and output generation.

Step	Name	What happens
6	Full Paper Download	Auto-retrieve PDFs from PMC, Unpaywall, Semantic Scholar
7	Full-Text Screening	AI reads entire papers and scores relevance section-by-section
8	Tabular Extraction	AI extracts structured data into custom-defined tables
9	Quality Assessment	AI-assisted RoB2 and GRADE with evidence citations
10	Meta-Analysis	Forest plots, effect sizes, risk of bias visualization
11	AI Writer	AI manuscript agent for drafting Results & Discussion
12	Export & Report	PRISMA flow diagram, BibTeX/RIS/CSV, activity log

Progress Tracking

The Workflow at a Glance sidebar on your project dashboard shows live stats:

Total papers in your library

Duplicates detected

AI Included / AI Excluded from screening

Finalized papers for analysis

The blue “Continue Your Review” banner always tells you the recommended next step. Follow it for the smoothest experience.

Validation Gates

EviSyn enforces quality through validation gates between stages:

You can’t start screening until papers are imported

You can’t finalize until human review conflicts are resolved

You can’t run quality assessment until papers are finalized

Each stage must be explicitly marked complete before the next unlocks

This prevents incomplete work from propagating through your review.

Identification & Screening