Practical Session: Data Extraction in RevProHub

Two-Phase Extraction 1‑to‑Many Repeater Logic 90 minutes

Practical Session Objectives:
By the end of this hands-on session, participants will be able to:

Navigate the RevProHub Data Extraction Dashboard
Assign extraction tasks to reviewers (bulk and individual)
Extract study characteristics and outcome data using structured forms
Use AI-assisted extraction from full text
Implement 1-to-many repeater fields for qualitative findings
Resolve extraction conflicts using the Mediator role
Export final consensus data for meta-analysis

Part 1: Data Extraction Dashboard

Navigate to: Main Project Dashboard → Conducting Phase → Data Extraction Dashboard

0
Ready to Extract

2
My In Progress

0
Pending Conflicts

0
Completed

N/A
Inter-Rater Reliability

1 Manage Forms: If no extraction form exists, the dashboard shows a setup alert. Create forms first in Planning → Extraction Form Planning (Session 2.6.12).

2 Assign All Pending: Use the modal to assign all awaiting studies to reviewers (select form, reviewers, start/due dates).

3 Studies Ready Table: Lists studies that passed screening, ready for extraction. Click "Extract" to begin.

Stats Cards Navigation: Click any card to navigate directly to the respective list (Ready Studies, My Extractions, Conflicts, Completed).

Part 2: Assigning Studies for Extraction

1 Click "Assign All Pending" button (or assign individually from Studies table).

2 Select Extraction Form (e.g., "Meta-Analysis of Effectiveness - Count Data").

3 Select Reviewer(s) – hold Ctrl/Cmd for multiple (enables dual extraction).

4 Set Start Date and Due Date (optional).

5 Click "Assign Queue" – reviewers receive email notifications and tasks appear in "My Extractions".

Dual extraction quality control: Assigning two reviewers to the same study enables automatic conflict detection and resolution, improving data accuracy.

Part 3: The Extraction Interface

Access via: My Extractions → Click study title → Extract with Form

AI Extraction Tools

Study Characteristics (Static – extracted once)

Study Author & Year

Study Design

Intervention Group Description

Intervention Group Sample Size (N1)

Intervention Group Events (n1)

Control Group Sample Size (N2)

Control Group Events (n2)

Control/Comparator Description

Outcome Measure Description

Follow-up Period

Notes

Qualitative Findings (1‑to‑Many Repeater Group)

Theme/Category Name

Illustrative Quote

Credibility Rating

Auto-save: Progress is auto-saved every 30 seconds. Use "Save Progress" button for manual save. Click "Submit Final Extraction" when complete.

Part 4: My Assigned Extractions

Navigate to: Data Extraction Dashboard → My In Progress card → My Extractions

Title	Authors	Year	Status	Actions
Metabolic dysfunction-associated steatotic liver disease (MASLD)...	Knorr JP et al.	2026	in progress
Evaluation of asthma control among adult patients...	Titus GJ et al.	2025	in progress

🔍 Bulk actions: Select multiple studies to clear records or unassign yourself from extractions.

Part 5: Extraction Conflict Resolution (Mediator Role)

Reviewer A:
Intervention N1: 120
Events: 45

Reviewer B:
Intervention N1: 125
Events: 48

Mediator Final:

Inter-Rater Reliability (Kappa): Dashboard calculates Cohen's Kappa for any selected extraction form when sufficient paired data exists.

Part 6: Import, Export & Data Cleaning

Bulk Assessment via CSV
Download template → Extract offline → Import back

Conflict Resolution via CSV
Export conflicts → Resolve in CSV → Re-import resolutions

Data Cleaning Utilities
Standardize dates, trim whitespace across all studies

📤 Export Final Consensus Data: Select form → Choose CSV or DOCX → Export resolved data for meta-analysis in R, SPSS, or RevMan.

Pro Tip for Qualitative Meta-Aggregation: When importing CSV data, include multiple rows for the same Study ID. The system automatically groups them as multiple findings under the repeater group.

Part 7: All Studies Ready for Extraction

Navigate from Dashboard → "Ready to Extract" card or "View List" button.

ID	Title	Authors	Year	Status	Actions
6002	Evaluation and determinants of asthma control among adult patients...	Titus GJ et al.	2025	Ready
5999	Metabolic dysfunction-associated steatotic liver disease (MASLD)...	Knorr JP et al.	2026	Ready

🔎 Features: Search by title/author/year, sortable columns, pagination (10/25/50/100 per page), and individual study assignment.

Part 8: Studies with Pending Conflicts

Navigate from Dashboard → "Pending Conflicts" card → View All.

If no conflicts exist, the list shows "No studies have pending conflicts." Conflicts arise when two reviewers extract different values for the same field.

🧪 Group Hands-On Activity: Data Extraction (50 min)

Access Extraction Dashboard: In your training project, go to Conducting Phase → Data Extraction Dashboard.
Create/Verify Extraction Form: Ensure you have a form from Session 2.6.12 (e.g., "Meta-Analysis of Effectiveness - Count Data" or "Qualitative Synthesis with Repeater").
Assign Studies (Project Manager role): Click "Assign All Pending" → select your form → choose two reviewers (e.g., two team members) → assign. Verify emails/tasks appear.
Extract Data (Reviewer A): Go to "My Extractions" → open a study. Fill Study Characteristics (Author, Year, N1, N2, events). For the repeater group (if qualitative form), add 2-3 theme rows with illustrative quotes. Click "Save Progress", then "Submit Final Extraction".
Extract Data (Reviewer B – same study): Repeat step 4 but introduce at least one discrepancy (e.g., different N1 or events).
Simulate Conflict: After both submit, check Dashboard → "Pending Conflicts" should show the study with number of conflicts.
Mediator Resolution: A third team member (Mediator role) navigates to Conflict Resolution page. Review side-by-side discrepancies, select final values, add rationale, and resolve each field.
Export Final Data: After resolution, go to "Manage Data" → Export Final Consensus Data → select form → export CSV. Open in Excel to view cleaned, reconciled dataset.
(Optional) AI Extraction Test: Paste a study abstract into the AI Extraction Tools textarea, click "AI Extract Data", and observe auto-populated fields.

Debrief Questions:

How does two-phase extraction (Study Characteristics + Outcome Data) improve data quality?
What is the advantage of the 1-to-many repeater logic for qualitative evidence synthesis?
How does the Mediator conflict resolution workflow reduce bias compared to senior reviewer overriding decisions?
When would you use CSV bulk import vs. online extraction?
How can Inter-Rater Reliability (Kappa) help monitor extraction quality across your team?

Connecting to Data Extraction Theory (Session 3.3)

Key Concepts Applied in RevProHub Extraction Module:

Two-Phase Approach: Study characteristics (static) separated from outcome data (repeatable) – aligns with Cochrane/JBI recommendations.
Standardized Forms: Prevents ad-hoc extraction, ensures consistency across reviewers.
Dual Extraction: Two independent reviewers + conflict resolution = reduced data entry errors.
1-to-Many Repeater Logic: Enables qualitative meta-aggregation (multiple findings per study).
AI Assistance: Speeds up extraction by auto-populating fields from full text (human verification still required).

Data Extraction Module Completion Checklist:
✔ Extraction form created in Planning Phase (Session 2.6.12)
✔ Studies assigned to reviewers (bulk or individual)
✔ Study characteristics extracted and saved
✔ Outcome data extracted (quantitative) OR themes/quotes extracted (qualitative with repeater)
✔ Conflicts resolved by Mediator
✔ Final consensus data exported for analysis
✔ Understanding of CSV bulk import for offline extraction

Next Practical Session: Synthesis & Reporting (Meta-analysis, PRISMA diagram, Manuscript Editor).