SignalsLab Review
A desktop application for systematic reviews and references. Screening, full text, extraction, risk of bias and synthesis on your own computer, inside a project folder you choose, with the references, the PDFs and the exports kept beside the review. Free software, AGPL-3.0.
How it works
What it adds to the review tools you already know.
Everything stays with you
The review is one file in a folder you choose, next to your raw database exports, the deduplicated library, the EndNote 21 library and every PDF. No account on anyone's platform, no per-project fee, and a trial mode that works alone from the first minute.
A reference manager inside the review
Import RIS, PubMed, EndNote XML or CSV from PubMed, Scopus, ScienceDirect, Web of Science, IEEE Xplore, Embase and the rest. The library reads like Zotero: collections, a table, an item pane with the abstract, attachments, tags and notes. Duplicates are found and verified by hand; the deduplicated library exports with PDF links Zotero can import.
Screen, read, extract, judge
Title & abstract with Yes / No / Maybe on the keyboard, full text with the PDF beside the decision and excerpts you tag as you read, extraction in sections with no blank fields and every value flagged extracted, calculated or imputed, risk of bias one study at a time in each tool's own words (RoB 2, ROBINS-I, QUADAS-2, PROBAST, Newcastle-Ottawa). Every exclusion carries one reason from an ordered list you set.
Study ids, Meta Lab, PRISMA 2020
Included studies get first-author-and-year ids that follow them into extraction, the bibliography (RIS, BibTeX, EndNote XML) and every export. Meta Lab pools the extracted estimates for a first look; PRISMA 2020 counts are assembled from the records, and the checklist is written into the project folder with what the review already knows.
Request a copy, or write to us
Installers are sent on request. For a copy for Windows, a macOS build when it is ready, a review you would like to run with us, or any question, use the request form: it opens a mail to us in your own mail program.
Why we built it
The usual review platforms keep the work on their own site and charge per project, and the references live somewhere else again. SignalsLab Review keeps the review and its references together on your own computer, as one file in a folder beside the raw exports and the PDFs, and works with the tools already in use: EndNote 21 for full texts, Zotero for the library, R for the statistics.
Three rules shaped it. Reviewers work independently and compare only afterwards. Every decision carries who made it, when, and under which version of the protocol, and nothing is ever deleted. And no number that reaches a manuscript comes from the application: the statistics are computed in R, and Meta Lab is for a first look.
Built on open work
What the application takes from existing open-source and open-standard efforts, and what each is used for.
PRISMA 2020 and the guides
The PRISMA 2020 flow is assembled from the record data rather than typed in, and the checklist is written with what the review already holds. The stages, the vote words, the ordered exclusion reasons, the conflict models and the risk-of-bias scales follow the published screening, extraction and review guides rather than our own conventions.
Zotero, EndNote 21, pdf.js
The library follows Zotero's design and exports what Zotero imports, PDF links included; EndNote 21 fetches the full texts and hands the library back; the reader is pdf.js, with excerpts tagged to the extraction fields and the risk-of-bias domains.
In the spirit of ASReview
A priority order for unscreened records, as ASReview does it, is part of the design: it orders, it never excludes, and work done with a suggestion in view is labelled as such. The ordering model is the next piece to land.
Meta Lab, then R
Meta Lab pools the extracted estimates by study id for a first look (inverse variance, random effects, the prediction interval shown with the pooled value). The numbers that reach a manuscript come from R with metafor, fed by the application's extraction export.
SQLite, one file per review
The review is a single SQLite file in your project folder: every record, vote, value, judgement, excerpt, tag and note, with the history of every change. Any SQLite tool opens it; every export is plain CSV, RIS, BibTeX or EndNote XML.