Screening

Screening is where most reviews stall. It doesn't have to be.

After the literature search, you're looking at thousands — sometimes tens of thousands — of records. Each one needs to be assessed against your inclusion criteria. PRISMA and Cochrane require independent dual-reviewer screening with documented conflict resolution. This is the step that turns a six-month review into an eighteen-month one.

mapped's screening workflow combines AI acceleration with methodological rigor to cut that timeline dramatically — without cutting corners.

---

Two-Phase Screening

Systematic reviews screen in two phases, and mapped supports both:

Phase 1: Title and Abstract Screening

The first pass. For each record, reviewers read the title and abstract and decide: include, exclude, or maybe. In a typical review, 90–95% of records are excluded at this stage.

mapped's AI Classification Engine pre-processes your queue in batches, flagging likely inclusions and exclusions based on your defined criteria. Reviewers then confirm or override each suggestion. This isn't blind automation — it's triage that lets you focus your attention where it matters.

Phase 2: Full-Text Screening

The second pass. Included records from Phase 1 are assessed in full text against detailed eligibility criteria. Exclusions at this stage require a documented reason (e.g., wrong population, wrong outcome, wrong study design).

mapped tracks exclusion reasons automatically and feeds them into your PRISMA flow diagram.

---

Dual-Reviewer Workflows

PRISMA and Cochrane guidelines require at least two independent reviewers for screening. mapped enforces this by design:

1. Assign reviewers — the project lead assigns reviewers to screening phases
2. Independent screening — each reviewer sees the same records but cannot see the other's decisions until both are complete
3. Automatic conflict detection — when reviewers disagree, the record is flagged for resolution
4. Conflict resolution — a third reviewer or the lead resolves disagreements with documented rationale
5. Agreement metrics — inter-rater reliability (Cohen's kappa) is calculated automatically

This isn't optional infrastructure — it's how the system works. You can't accidentally skip dual review.

---

AI Batch Processing

For large screening queues (5,000+ records), manual screening alone takes weeks. mapped's Classification Engine accelerates this:

• Batch pre-screening — AI processes records in bulk, assigning preliminary include/exclude/uncertain labels
• Confidence scoring — each AI decision comes with a confidence level, so you know which ones need closer human review
• Learning from your decisions — as you confirm or override AI suggestions, the system adapts to your specific inclusion criteria
• Priority ordering — uncertain records are surfaced first, so human attention goes where it adds the most value

The AI handles the obvious exclusions (wrong species, wrong study type, conference abstracts). You handle the borderline cases that require methodological judgment.

---

Real-Time Progress Tracking

For team-based reviews, knowing where you stand matters. mapped provides:

• Live progress dashboard — see how many records are screened, pending, and in conflict
• Per-reviewer statistics — track individual reviewer progress and screening speed
• Deadline projections — estimated completion dates based on current pace
• Stage-level visibility — see Phase 1 and Phase 2 status independently

---

Screening Data Export

All screening decisions are exportable for:

• PRISMA flow diagram — automatically populated with numbers at each stage
• Supplementary materials — full list of excluded studies with reasons
• Audit trail — every decision, by whom, at what time, with any AI assistance noted

---

Why This Step Matters

Screening quality determines review quality. Include a study that shouldn't be there, or miss one that should — and your conclusions are compromised. mapped's dual-reviewer enforcement, AI assistance, and conflict resolution ensure that screening is both thorough and efficient. The methodology is non-negotiable; the speed is dramatically improved.

---

Next step: With your included studies finalized, move to Data Extraction →