A free systematic review screening tool for title and abstract screening
This systematic review screening tool handles the study-selection stage of a review: title and abstract screening. You import your de-duplicated search results, define your eligibility concepts, and the tool ranks every record by relevance so the studies most likely to meet your criteria appear first. Screening this way is faster than reading records in arbitrary order, and because the tool never hides borderline records, recall is preserved: ranking orders the list, a human reviewer still decides every include and exclude. It is a free, account-free alternative to Covidence and Rayyan for the screening step. By default everything runs locally in your browser, so no record, abstract, or decision is uploaded to a server. When you want a second reviewer, an optional real-time collaboration mode lets you share one invite link and screen together; only then is the project synced, and only between people who hold the link.
The tool reads the formats databases and reference managers actually export, RIS, PubMed .nbib/MEDLINE, CSV, TSV, and Excel, and merges multiple files while removing duplicates by DOI, exact title, and fuzzy near-duplicate matching (Jaccard title similarity) that catches the same study indexed differently across databases. Ranking runs in a background worker and projects persist in your browser's database rather than its size-capped local storage, so reviews with tens of thousands of records stay responsive. As you mark records Include, Maybe, or Exclude, the optional active-learning classifier trains on your own decisions, re-ranks the unscreened pile, and estimates how many relevant records likely remain, giving you a defensible, recall-oriented stopping signal rather than a guess. Structured exclusion reasons (wrong population, intervention, comparator, outcome, or study design) are tallied automatically for your reporting. Because best practice is two independent reviewers, the Reconcile step loads two saved projects, scores agreement with Cohen's kappa, and lists every conflict to resolve, the offline equivalent of dual screening. You can map your question first with the PICO framework builder so the eligibility concepts you screen against are well defined.
Screening sits in the middle of the review pipeline. Before you screen, finalize your search with the search strategy builder and remove duplicates with the reference deduplication tool. Pin down your eligibility rules with the inclusion and exclusion criteria builder. When screening is done, the counts here flow straight into the PRISMA flow diagram generator for a publication-ready figure that meets PRISMA 2020 reporting standards expected by Cochrane, JBI, and major journals.
A free alternative to Rayyan and Covidence for screening
Most reviewers reach the screening stage already weighing a Rayyan alternative or a Covidence alternative, usually because of cost, a seat limit, or a data-residency policy that forbids uploading unpublished records to a third-party server. The difference here is where the work happens. Rayyan and Covidence are hosted platforms: your library, decisions, and reviewer activity live on their servers behind a subscription or a per-review charge. This tool keeps the same screening workflow, relevance ranking, structured exclusion reasons, two-reviewer reconciliation with Cohen's kappa, and PRISMA counts, but runs in your browser with nothing uploaded by default and no account required to start. There is no record cap and no paywall on the screening step itself.
The same reasoning applies to the wider category of systematic review software: DistillerSR and Covidence are paid hosted platforms, while ASReview and Abstrackr are free machine-learning screeners that still expect you to install software or upload your library. This tool sits alongside them as free, browser-based literature review software for the screening step, with active-learning prioritization built in and nothing to install.
It is not a like-for-like replacement for every team. Covidence bundles full-text review and data extraction, and Rayyan's paid tiers add team management; if you need a hosted shared workspace with managed accounts, those remain reasonable choices. What this tool replaces is the common case: a single reviewer or a small pair who need rigorous, recall-oriented title and abstract screening without a subscription, and who would rather keep unpublished search results on their own machine. When you do want a second reviewer, optional collaboration syncs the project only between people holding the invite link.
Systematic review screening software that runs in your browser
As systematic review screening software, this tool is local-first rather than cloud-first. There is nothing to install and no server round-trip on import or ranking: your file is parsed in the page, projects are stored in your browser's IndexedDB, and the active-learning classifier trains on-device, so a review with tens of thousands of records stays responsive offline. That design is what makes it genuinely private, the records, abstracts, and include/exclude decisions never leave your computer unless you deliberately turn on collaboration.
Because the engine is the same one a hosted platform would run, the outputs are publication-grade: a recall estimate for a defensible stopping point, a kappa score for inter-reviewer agreement, and exact counts that drop into a PRISMA 2020 flow diagram. The trade-off of local-first software is that your projects live in one browser profile, so export a backup if you switch machines, or sign in to sync saved reviews across devices.
How to screen with this tool
- 1
Import your search results
Upload or paste your search export in RIS, PubMed .nbib/MEDLINE, CSV, TSV, or Excel format. Multiple files merge and de-duplicate automatically, and the duplicate count feeds straight into your PRISMA identification numbers.
- 2
Set your relevance criteria
Enter inclusion concepts (one per line, commas for synonyms) and any exclude signals, then rank. Likely-relevant records rise to the top and exclude signals are highlighted in red but never auto-removed.
- 3
Decide and let the tool learn
Mark each record Include, Maybe, or Exclude with a structured reason, using keyboard shortcuts for speed. After at least one include and one exclude, the active-learning classifier re-ranks the rest and estimates how many relevant records likely remain so you know when it is safe to stop.
- 4
Reconcile two reviewers
Screen as two independent reviewers two ways. Offline: each reviewer saves a project file and you load both into the Reconcile step to score agreement with Cohen's kappa and resolve every conflict. Or opt into live collaboration: share one invite link, screen the same records together in real time, and watch agreement build as you go, the equivalent of dual screening in Covidence or Rayyan.
- 5
Export and chart
Download decisions as CSV, save the whole project as JSON to resume later, copy your PRISMA counts, or build a PRISMA 2020 flow diagram and export it as an image. Nothing is uploaded to a server.
Frequently asked questions
What is title and abstract screening?
Title and abstract screening is the first study-selection stage of a systematic review. After de-duplicating your database search results, two reviewers independently read each record's title and abstract and decide whether it could meet the eligibility criteria. The goal is high recall, not precision: a record is only excluded when its title or abstract clearly shows it does not fit the population, intervention, comparator, outcome, or study design (PICOS). Anything borderline is carried forward to full-text screening. This tool ranks records by relevance so likely-includes surface first, but a human reviewer still makes every include or exclude call.
How to do abstract screening?
Import your search export (RIS, PubMed .nbib/MEDLINE, CSV, TSV, or Excel) and de-duplicate. Define your inclusion concepts and exclude signals against your protocol. Then read each title and abstract and mark it Include, Maybe, or Exclude, recording a structured exclusion reason (wrong population, wrong intervention, wrong outcome, wrong study design, and so on). Screen liberally: keep anything you cannot confidently exclude. Best practice is two independent reviewers who screen the same records and then reconcile disagreements. This tool orders records by relevance, lets a classifier learn from your decisions, and exports the counts you need for the PRISMA 2020 flow diagram.
How to do title and abstract screening in Covidence?
In Covidence you import references, the platform removes duplicates, and two reviewers vote Yes / No / Maybe on each title and abstract; conflicts are resolved by a third reviewer or by discussion before full-text review. This browser-based tool follows the same two-reviewer workflow without an account or subscription: each reviewer screens in their own browser and saves a project file, then you load both files into the Reconcile step, which scores agreement with Cohen's kappa and lists every conflict to resolve. It adds relevance ranking and active-learning prioritization on top, and nothing ever leaves your browser.
What is study selection in systematic review?
Study selection is the process of deciding which records identified by your search are eligible for inclusion. It runs in two stages: title-and-abstract screening (a fast first pass to remove clearly irrelevant records) followed by full-text screening (a detailed eligibility check of the remaining reports). Each exclusion at the full-text stage must be recorded with a reason. The numbers from both stages, identified, duplicates removed, screened, excluded, and included, populate the PRISMA 2020 flow diagram and must be reported transparently so the review is reproducible.
What are the 7 steps of a systematic review?
A systematic review typically follows seven steps: (1) formulate a focused, answerable question (often using PICO); (2) write and register a protocol (for example on PROSPERO); (3) run a comprehensive, reproducible search across multiple databases; (4) screen records by title and abstract, then full text, against eligibility criteria; (5) extract data from included studies; (6) assess risk of bias and certainty of evidence (for example with RoB 2 and GRADE); and (7) synthesize the findings, narratively or through meta-analysis, and report following PRISMA 2020. This tool supports step 4, the screening and study-selection stage.
Screening hundreds or thousands of records and want a second methodologist on the calls? Get a quote for screening support from our PhD methodologists.