User Guide
How to read and use validation reports
Quick Start
1. Find a Drug of Interest
- Use the search function in the top right corner
- Or browse from the Drug List
- Or filter by Evidence Level
2. Check Evidence Level
Each drug has an evidence level (L1-L5):
| Level | Meaning | Confidence |
|---|---|---|
| L1 | Multiple large RCTs support | ⭐⭐⭐⭐⭐ |
| L2 | Single RCT or multiple Phase 2 | ⭐⭐⭐⭐ |
| L3 | Observational studies | ⭐⭐⭐ |
| L4 | Preclinical/mechanistic studies | ⭐⭐ |
| L5 | AI prediction only | ⭐ |
3. Read the Full Report
Reports contain detailed evidence analysis and recommendations.
Report Structure
Each report contains the following sections:
One-Sentence Summary
Quickly understand the core content:
- Original indication
- Predicted new indication
- How much evidence supports it
Quick Overview Table
| Field | Description |
|---|---|
| Original Indication | Drug's currently approved use |
| Predicted New Indication | TxGNN's predicted new use |
| TxGNN Prediction Score | AI confidence level (higher = more confident) |
| Evidence Level | L1-L5, see above |
| Singapore Registration | Whether registered with HSA |
| Recommendation | Go / Proceed / Consider / Explore / Hold |
Why This Prediction Makes Sense
Explains the pharmacological mechanism:
- Drug's mechanism of action
- Why this mechanism might work for the new indication
- Relationship with original indication
Clinical Trial Evidence
Lists relevant clinical trials:
| Field | Description |
|---|---|
| Trial ID | ClinicalTrials.gov ID, clickable |
| Phase | Phase 1/2/3/4 |
| Status | Ongoing, Completed, Terminated, etc. |
| Enrollment | Number of participants |
| Key Findings | Trial results (if available) |
Literature Evidence
Lists relevant academic literature:
| Field | Description |
|---|---|
| PMID | PubMed ID, clickable |
| Year | Publication year |
| Type | Article type (journal article, review, etc.) |
| Journal | Publication journal |
| Key Findings | Main findings |
Singapore Registration Info
Lists the drug's HSA registration:
- License number
- Product name (brand name)
- Dosage form
- Approved indications
Safety Considerations
Highlights potential safety issues.
Conclusion & Next Steps
Summarizes the recommendation and explains:
- Why this recommendation was made
- What's needed to proceed
Decision Recommendations
| Decision | Meaning | Suggested Action |
|---|---|---|
| Go | Very strong evidence | Can proceed to detailed evaluation or trial planning |
| Proceed | Strong evidence | Worth further feasibility assessment |
| Consider | Some evidence | Can consider, but weigh risks |
| Explore | Worth exploring | Recommend gathering more data first |
| Hold | Insufficient evidence | Not recommended to proceed currently |
Frequently Asked Questions
Q: Can I use L1 drugs directly?
Not necessarily. L1 indicates strong clinical evidence supporting the prediction, but:
- Still need to confirm regulatory compliance
- Need to assess individual applicability
- Need to consider drug interactions
Q: Are L5 drugs useless?
Not necessarily. L5 indicates lack of clinical evidence currently, but:
- May be an unexplored area
- AI predictions can serve as research hypothesis starting points
- Suitable for early research directions
Q: Is a higher prediction score always better?
Generally yes. Prediction score reflects AI model confidence:
- Above 99.9%: Very confident
- 99.0-99.9%: Confident
- But high prediction score ≠ high clinical evidence
Q: How often are reports updated?
Reports are currently generated once reflecting data at generation time. Future regular updates may be implemented.
Q: Can this be used clinically?
Absolutely not. This report is for research reference only and does not constitute medical advice. Any clinical application requires:
- Complete clinical trial validation
- Regulatory approval
- Professional medical judgment
Still Have Questions?
Please ask through GitHub Issues.