Protocol Builder
The Protocol Builder is EvidAI's intelligent assistant for creating rigorous, PRISMA-compliant research protocols. It guides you through every methodological decision with AI-powered suggestions, ensuring your protocol meets the highest standards for systematic reviews.
A well-designed protocol is the foundation of any successful systematic review. It pre-specifies your methods, reduces the risk of bias, and is often required for journal publication and registration with PROSPERO (the International Prospective Register of Systematic Reviews).
Why Protocols Matter
Before diving into the Protocol Builder workflow, it's important to understand why investing time in protocol development pays dividends throughout your review:
| Benefit | Description |
|---|---|
| Reduces Bias | Pre-specifying methods prevents post-hoc decisions that could introduce bias |
| Ensures Reproducibility | Other researchers can replicate your methodology exactly |
| Facilitates Registration | PROSPERO and other registries require detailed protocols |
| Strengthens Publications | Journals increasingly require registered protocols for systematic reviews |
| Improves Team Alignment | Everyone knows the methodology before starting work |
Did you know? Studies with registered protocols are cited more frequently and are perceived as more credible by the research community.
The Protocol Builder Workflow
The Protocol Builder walks you through a structured, six-step process. At each step, AI assistance helps you make informed decisions.
Step 1: Define Your Research Question
Every systematic review begins with a well-formed research question. The Protocol Builder helps you structure your question using established frameworks.
The PICO Framework
PICO is the most widely used framework for clinical questions:
| Element | Question | Example |
|---|---|---|
| P — Population | Who are you studying? | Adults aged 65+ with Type 2 diabetes |
| I — Intervention | What treatment or exposure? | SGLT2 inhibitors |
| C — Comparator | What is the alternative? | Standard care or placebo |
| O — Outcome | What results matter? | All-cause mortality, cardiovascular events |
Alternative Frameworks
Depending on your research question, other frameworks may be more appropriate:
- PICOS — Adds Study design (for methodology-specific questions)
- PICOT — Adds Timing (for time-dependent outcomes)
- SPIDER — Sample, Phenomenon, Design, Evaluation, Research type (for qualitative reviews)
- PCC — Population, Concept, Context (for scoping reviews)
The AI analyzes your question and recommends the most suitable framework automatically.
Step 2: Methodology Selection
Based on your structured question, the Protocol Builder recommends the most appropriate methodology:
| Review Type | Best For | Typical Timeline |
|---|---|---|
| Systematic Review | Comprehensive evidence synthesis | 6-18 months |
| Rapid Review | Time-sensitive decisions | 2-6 months |
| Scoping Review | Mapping a field, identifying gaps | 3-9 months |
| Living Review | Topics with emerging evidence | Ongoing |
For each methodology, you'll also receive recommendations for:
- Which study designs to include (RCTs, cohort studies, etc.)
- Appropriate quality assessment tools (RoB 2, ROBINS-I, etc.)
- Synthesis approaches (narrative, meta-analysis, mixed methods)
Step 3: Search Strategy Development
A comprehensive search strategy is critical for a valid systematic review. The Protocol Builder helps you develop yours:
Database Selection — Based on your topic, the AI recommends which databases to search. For medical reviews, this typically includes PubMed/MEDLINE, Embase, CENTRAL, and specialized databases.
Concept Blocks — Your PICO elements become the foundation for search concept blocks. The AI helps expand each concept with synonyms, MeSH terms, and related terms.
Boolean Structure — The system generates properly structured Boolean searches with appropriate use of AND/OR operators, truncation, and proximity searching.
Example: For a PICO question about "SGLT2 inhibitors for heart failure," the AI might suggest searching for: (SGLT2 inhibitors OR sodium-glucose cotransporter 2 inhibitors OR empagliflozin OR dapagliflozin OR canagliflozin) AND (heart failure OR cardiac failure OR HFrEF OR HFpEF)
Step 4: Eligibility Criteria
Clear, testable eligibility criteria prevent ambiguity during screening. The Protocol Builder helps you define:
- Inclusion criteria aligned with your PICO elements
- Exclusion criteria that filter out irrelevant studies
- Pilot-testable definitions that multiple screeners can apply consistently
The AI also warns about common pitfalls, such as criteria that are too restrictive (missing relevant studies) or too broad (overwhelming screening burden).
Step 5: Data Extraction Planning
Pre-planning your data extraction ensures you capture all necessary information:
- Standard fields — Author, year, study design, sample size, etc.
- PICO-aligned variables — Population characteristics, intervention details, comparator specifics, outcome measures
- Risk of bias domains — Items needed for quality assessment
- Synthesis variables — Effect sizes, confidence intervals, heterogeneity measures
The system generates customizable extraction forms that your team can use directly in Evidence Studio.
Step 6: Analysis Planning
Document your synthesis approach before beginning the review:
- Qualitative synthesis — How will you narratively summarize findings?
- Quantitative synthesis — Is meta-analysis feasible and appropriate?
- Subgroup analyses — What pre-specified subgroups will you examine?
- Sensitivity analyses — How will you test the robustness of your findings?
AI-Powered Features
Question Refinement
The AI analyzes your research question for:
| Criterion | What the AI Checks |
|---|---|
| Specificity | Is the question focused enough to be answerable? |
| Feasibility | Is there likely enough evidence to synthesize? |
| Alignment | Does the methodology match the question type? |
| Scope | Is the question too broad or too narrow? |
Methodology Recommendations
Based on your question, the AI provides tailored recommendations for review methodology, search databases, quality assessment tools, and statistical approaches.
Protocol Templates
Jump-start your protocol with pre-built templates for common review types:
- Clinical intervention reviews
- Diagnostic accuracy reviews
- Prognostic factor reviews
- Public health intervention reviews
PROSPERO Integration
Registration-Ready Export
Your completed protocol can be exported directly for PROSPERO submission. The system automatically formats all required fields and generates the XML structure needed for registration.
What's Included
| Field Category | Auto-Generated Content |
|---|---|
| Review Information | Title, anticipated start/completion dates |
| Team Details | All authors with roles and affiliations |
| Methods | Full methodology including PICO, search strategy, eligibility criteria |
| Analysis | Synthesis plans, subgroup analyses, risk of bias approach |
Export Options
Document Formats
Export your protocol in the format you need:
- Word (.docx) — Fully formatted, editable document
- PDF — Print-ready, fixed-layout version
- Markdown — Plain text with formatting for version control
- HTML — Web-publishable format
Structured Formats
For registration and compliance:
- PROSPERO XML — Direct submission format
- PRISMA-P Checklist — Completed protocol checklist
- Journal Templates — Pre-formatted for specific journals
Best Practices
Before You Start
- Conduct preliminary searches — Understand what evidence exists
- Check for existing reviews — Avoid duplicating recent work
- Assemble your team — Identify co-reviewers, methodologists, and content experts
- Allocate sufficient time — Protocol development typically takes 2-4 weeks
During Protocol Development
- Be specific but not restrictive — Criteria should be clear but not exclude important evidence
- Plan for amendments — Document a process for protocol changes if needed
- Document all decisions — Keep a rationale log for methodological choices
- Get peer feedback — Have colleagues or methodologists review your protocol