DOCSEARCHbyAlgolia
Submit Docs

abstrackr Documentation

website·Indexed 32 days ago

https://docs.abstrackr.com

Documentation for abstrackr, a tool for systematic and scoping reviews. It features tools for abstract screening, including single and double screening modes, pilot rounds, conflict resolution, and PRISMA flow diagram counts. The platform supports importing various citation file formats, machine learning-based probability scoring, and exporting data via CSV, RIS, and JSONL.

Tokens
19K
Snippets
26
Records
161
Agent score
49%

What's inside abstrackr

  1. Understand the purpose of Screening Sets

    A Screening Set is a fixed, named collection of citations used to organize and assign screening work. Unlike live filtered lists—where labeling a citation causes it to disappear and shift the user's position—a set freezes the list. Citations that are labeled are marked as done but remain visible, ensuring the user's position in the list never changes.

  2. Understand Blinding in Abstrackr

    Blinding is a per-project setting that prevents reviewers from seeing each other's decisions, tags, and notes during the screening process. This ensures that each reviewer forms an independent judgment without being influenced by colleagues. Blinding is enabled by default when a project is created and is managed by project leaders.

  3. Manage keyword highlighting with Personal and Project terms

    abstrackr allows reviewers to highlight keywords in citation titles and abstracts to speed up the screening process. Terms are divided into two scopes:

    • Personal Terms: Set by individual reviewers. These are private to the user and apply across all projects they participate in.
    • Project Terms: Set by project leaders. These are shared with all team members and are typically used for core inclusion/exclusion criteria.

  4. Choose a screening mode for citation review

    The screening mode determines how many independent reviewer decisions are required before a citation is considered resolved.

    • Single Mode: Requires 1 reviewer. Best for rapid screening or expert-led reviews.
    • Double Mode: Requires 2 reviewers. Best for rigorous systematic reviews.
    • Pilot Mode: Requires all team members to label every citation. Best for the calibration phase at the start of a project (typically the first 50–100 citations) to ensure inclusion criteria are understood consistently.

  5. Understand screening sets in abstrackr

    A screening set is a named, immutable snapshot of citation IDs captured at the time of creation. Unlike the live open-queue, which shifts as citations are labeled, a set freezes the list of citations based on the filters and sort order active during its creation. Labels are global and citation-level; labeling a citation in one set applies that label across the entire project, including other sets containing the same citation.

  6. Understand global label behavior in screening sets

    Labels in abstrackr are global. When a reviewer labels a citation in one screening set, that label is recorded project-wide. If the same citation appears in another set, it is automatically marked as done in that set as well, preventing duplicate work across different sets.

  7. Use abstrackr for abstract screening workflows

    abstrackr is designed to manage the abstract screening stage of a systematic review. Its core capabilities include:

    • Importing citations from sources like PubMed, Embase, and Zotero.
    • Providing an interface for Include/Exclude/Maybe decisions.
    • Supporting blinded independent screening to maintain reviewer independence.
    • Tracking team progress and managing the conflict resolution process.
    • Automatically generating PRISMA flow numbers based on screening decisions.
    • Exporting results for the full-text eligibility stage.

  8. Understand the abstract screening process in systematic reviews

    Abstract screening is a critical stage in a systematic review where researchers evaluate thousands of database search results to determine which studies are worth reading in full. The goal is to filter out irrelevant records based only on their titles and abstracts before proceeding to full-text retrieval. This process follows the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) framework, specifically the 'Screening' stage, which sits between 'Identification' (retrieving records) and 'Eligibility' (assessing full texts).

  9. Define inclusion criteria using the PICO framework

    For clinical systematic reviews, structure inclusion criteria using the PICO framework to ensure all eligible studies address four key elements. For non-clinical reviews, adapt these to Population/Phenomenon, Exposure/Concept, Context, and Outcome.

    ElementQuestion it answersExample
    PopulationWho are the participants?Adults aged ≥18 with a confirmed diagnosis of Type 2 diabetes
    InterventionWhat was done or given?Metformin monotherapy, any dose
    ComparatorWhat was it compared to?Placebo, diet/exercise alone, or another anti-diabetic drug
    OutcomeWhat was measured?HbA1c reduction at 12 months or longer

  10. Understand ML model activation and citation sorting

    abstrackr's machine learning model assigns a probability score (0–1) to every citation to estimate relevance. The model runs approximately weekly and updates scores as more citations are screened.

    Activation Requirement: The model requires approximately 100 labels (a mix of included and excluded decisions) before it generates meaningful scores. Before this threshold is met, citations are displayed in import order.

    Sorting Behavior: Once ML scores are available, the screening queue defaults to 'ML Priority order' (highest probability first). Users can manually switch back to chronological order via the screening page settings.

  11. Manage project exports while Blinding is active

    Blinding restricts the data available in project exports to prevent unauthorized access to other reviewers' decisions:

    • CSV / RIS export: Only the current user's labels, tags, and notes are exported. The resulting filename includes "(blinded)".
    • JSONL full project export: The download button is disabled while blinding is active. The project must be unblinded before a full JSONL export can be performed.

  12. Create a solo screening project in abstrackr

    To conduct a review where one person performs all screening (e.g., scoping or rapid reviews), create a project in 'Single' mode. This ensures each citation is resolved by a single decision without requiring a second reviewer.
    1. Sign in at abstrackr.com and click New Project.
    2. Enter project name and description.
    3. Set Screening Mode to Single.
    4. Click Create Project.