Run title, abstract, and full-text screening with reviewer control

The complete AccuScreener workflow

Screening moves from project setup to first-pass review, then to PDF preparation and full-text review. Later stages depend on the configuration and decisions made earlier, so changes should be reviewed before relying on exports or downstream handoffs.

1. 1

   Create or link the project

   Start a standalone screening project or open the screening stage inside a master project workflow.

2. 2

   Configure protocol, criteria, and team

   Set project details, select screening type, enable AI mode if needed, assign analysts, select the eligible reviewer pool, and confirm eligibility categories.

3. 3

   Add studies and resolve duplicates

   Load records from AccuSearch or upload Excel, RIS, or NBIB files, then apply automatic or manual duplicate review.

4. 4

   Complete first-pass screening

   Screen titles and abstracts manually, with AI suggestions, or against an independent AI Screener decision stream.

5. 5

   Prepare full-text review

   Upload or auto-download PDFs, match files to studies, resolve text extraction status, publish FTR configuration, and clear readiness blockers.

6. 6

   Review, adjudicate, and export

   Complete FTR, resolve required FPS and FTR adjudication, inspect audit history, and export screening decisions, PRISMA data, and timeline records.

Choose how AI participates before screening begins

AccuScreener supports manual screening, visible AI assistance, and independent AI screening. The mode changes what screeners can see, how conflicts are created, and how reviewers adjudicate final eligibility.

Assign analysts, reviewers, and project-level responsibility

Project setup separates screening work from adjudication work. Analysts screen assigned records, while the selected reviewers form an eligibility pool. Reviewer work is not distributed to every study during setup; it is created later only when a completed FPS or FTR outcome requires adjudication.

Configure the screening rules that drive decisions

AccuScreener uses protocol data, publication types, and inclusion or exclusion categories to guide first pass screening, full-text review, AI runs, reviewer queues, and exports. These settings should match the approved review protocol before screening starts.

Add studies from AccuSearch or supported uploads

A screening project can receive records from a linked AccuSearch workflow or from uploaded study files. Confirm the source set before screening so analysts and AI process the intended records.

Excel upload expectations

• Required columns include ID, Author, Year, Title, Journal, Abstract, DOI, and Article type.
• Study IDs should be positive whole numbers so decisions and exports remain traceable.
• RIS and NBIB imports can assign project IDs while retaining source identifiers.
• Upload previews should be checked before adding records to the project.

Resolve duplicate records before they distort screening

Duplicate detection keeps multiple citations for the same study from creating redundant work or conflicting decisions. AccuScreener supports automatic duplicate marking and manual duplicate review.

Complete first-pass screening with clear decision context

First-pass screening uses study titles, abstracts, publication information, protocol data, and configured categories to capture Include, Exclude, or Unsure decisions. The visible information changes by role, assignment, project mode, and whether AI Assistant or AI Screener is active.

Use Advanced Screening for controlled bulk decisions

Advanced Screening is an assignment-scoped FPS worklist for finding studies and applying the same decision to multiple selected records. It does not provide access to studies that are not assigned to the current screener or reviewer.

Use AI runs as auditable screening operations

AI Runs create traceable batches for first-pass screening or full-text review. Each run is tied to the selected protocol and, for FTR, the selected FTR configuration, so the decision context can be reviewed later.

Prepare PDFs before enabling full-text review

Full-text review depends on eligible first-pass decisions, matched PDFs or explicit PDF skips, text extraction, and a published FTR configuration. The PDF Manager helps teams upload, match, retry, skip, and audit files before FTR begins.

Use FTR configuration, PDFs, and reviewer decisions together

Full-text review combines the study list, PDF viewer, review questions, Mini-DET fields, and decision dialogs in one workspace. FTR evaluates its reviewer requirement separately from FPS. Decisions are associated with protocol and FTR configuration versions so later exports can show the rules used for the work.

Reviewer work is assigned when a study needs it

Selecting reviewers during project creation defines who is eligible to adjudicate; it does not assign every study to a reviewer. After the required analyst or AI screening decisions are saved, AccuScreener evaluates that study and either finalizes an accepted outcome or creates one reviewer task.

AI Assistant provides advice, not an independent screening arm. Follow the applicable human single or double review rule. Sole AI Screener follows the separate rules below.

How AccuScreener selects the reviewer

1. Wait until all required screening arms for the phase have submitted their decisions.
2. Classify the outcome as no adjudication required or review required.
3. Start with the active reviewers selected for the project and remove anyone who screened that study.
4. Assign the eligible independent reviewer with the lowest unresolved workload. Ties use a stable rotation so the result is predictable.
5. If nobody is eligible, save the screening work and mark the study Review required, unassigned for administrator action.

Examples

FPS and FTR are separate reviewer lifecycles

An FPS reviewer assignment is not copied into FTR. Unresolved FPS Unsure and decision conflicts block ordinary FTR. If both FPS arms agree on Include but their criteria differ, the study can enter FTR while its FPS criteria review remains pending. When FTR screening is later completed, AccuScreener evaluates the FTR decisions independently and creates FTR reviewer work only if the FTR outcome requires it.

Reviewer statistics and worklists count actual review work. Administration statistics separately identify outcomes finalized without adjudication and studies that require review but do not yet have an eligible reviewer. Duplicate secondary records are handled through the duplicate workflow and are not distributed as ordinary reviewer tasks.

Review quality across the project

Assign master reviewers for FPS, FTR, or both during team setup or in Change Team. Master review becomes available after screening is complete for the sole screener or one full arm in a double-blind project. Master decisions take precedence for their phase and retain an audit history.

The dedicated walkthrough explains All studies and non-overlapping random sets, Reading and Excel-style views, filters, bulk decisions, and what happens to FTR when a master reviewer excludes FPS.

Definitive AI decisions can finalize while human review remains available

Sole AI Screener is a single review workflow. A definitive AI Include or Exclude becomes the final FPS decision immediately, so an included study can proceed toward FTR without waiting for every record to be reviewed. AI Unsure and failed or unfinished AI processing remain blocked. Selected project reviewers can optionally inspect definitive AI decisions; when a reviewer submits a decision, that human decision becomes final and can verify or override the AI result.

FTR has its own Optional AI review queue for definitive, unreviewed AI FTR Include or Exclude results on FTR-eligible studies. Selected independent reviewers can verify or override those results without waiting for a required assignment. The ordinary FTR All assigned reviews list continues to show only the reviewer's assigned or completed work; pending AI results, Unsure outcomes, duplicates, and studies requiring adjudication do not enter the optional queue.

AI Reviewer FPS Statistics

Sole AI Reviewer replaces the ordinary FPS reviewer table with a sole-AI-specific table. It separates required resolution work from optional quality-assurance samples and shows whether completed human reviews agreed with or changed the AI result.

When adjudication is required

Once the required screening decisions are complete, Unsure outcomes and decision or criteria conflicts create reviewer work. The same classifications apply whether Arm A is screened by AI or a person.

The reviewer dropdown groups unfinished work by its reason: Unsure to resolve, Decision conflict, or Criteria conflict. All assigned reviews includes unfinished and completed work, with unfinished studies first; it also retains any legacy task without a saved reason. FPS additionally has My Unsure decisions for Unsure outcomes submitted by the reviewer. FTR reviewers must choose Include or Exclude; an unresolved study should be left open, not finalized as Unsure. If no independent reviewer is eligible, the study is marked Review required, unassigned instead of entering anyone's worklist.

Know when earlier work needs review

Screening is cumulative. Changes to the source set, protocol, FTR configuration, team assignment, AI runs, or first-pass decisions can affect downstream queues and outputs.

Use outputs with their decision context

AccuScreener exports and timeline views are designed to show not only the final decision, but also the setup, reviewers, AI runs, duplicate handling, PDF status, and history behind that decision.

How PRISMA chooses inclusion and exclusion reasons

PRISMA uses the saved final decision to count included and excluded studies. For each study, it selects one reason for the reported stage and outcome: FPS exclusion, FTR exclusion, or FTR inclusion. An eligible reason must be non-empty and belong to a completed decision matching that final outcome.

If both arms exclude with different reasons and no eligible reviewer reason is available, Arm B's reason is used. The same priority applies to FTR inclusion reasons. If Arm B has no eligible reason, Arm A is considered. For an agreed Include with an FPS criteria conflict, a pending reviewer does not supply a reason, so Arm B's matching reason is preferred where that FPS reason is reported. A completed reviewer reason takes priority later. The order in which analysts submit decisions does not determine arm priority.

• Pending or unresolved final decisions do not contribute inclusion or exclusion reasons.
• If a final outcome has no eligible reason, the study is counted under “Reason not recorded”.
• Each study contributes only once to its stage's reason breakdown; reasons from both arms are not added together.
• FPS inclusion totals also include Unsure; FTR inclusion totals require an Include outcome.

Use AccuScreener as a governed screening system

Final checklist before using results downstream

• check_circleThe active protocol, publication types, and decision categories match the approved review plan.
• check_circleAnalysts, the eligible reviewer pool, review type, and AI mode are configured intentionally.
• check_circleStudy records were imported from the intended AccuSearch source or validated upload file.
• check_circleDuplicate detection has been run and unresolved groups have been reviewed.
• check_circleFirst-pass decisions are complete; Unsure and decision conflicts are adjudicated before FTR. Agreed Include with a criteria-only conflict may move while its FPS review remains pending.
• check_circlePDFs are matched, skipped, or explicitly handled, and extraction blockers have been reviewed.
• check_circleFTR configuration is published and tied to the intended protocol version.
• check_circleAI re-screening, protocol changes, and FPS-to-FTR eligibility reviews have been addressed before export.

Common AccuScreener questions