How to Read Validation Reports

Each report has three key pieces of information: Evidence Level (L1-L5), Prediction Score (AI confidence), and Decision Recommendation (Go/Proceed/Hold). These three combined quickly tell you the prediction's reliability and recommended action.

Report Structure Overview

1. One-Sentence Summary

The one-sentence summary at the beginning is the most important quick-read section:

Example: Prednisolone is originally used for inflammatory conditions. TxGNN predicts it may be effective for 3,650 additional indications, with clinical trial evidence supporting some predictions.

This tells you:

The drug’s original use
The AI’s predicted new use
How much evidence supports it

2. Quick Overview Table

Item	How to Interpret
TxGNN Prediction Score	99%+ indicates very high AI confidence
Evidence Level	L1 is strongest, L5 is prediction only
Decision Recommendation	Go = proceed directly, Hold = wait
NPRA Status	Whether approved in Malaysia

3. Clinical Trial Evidence

Clinical trials are the most important evidence source:

Phase	Description	Evidence Strength
Phase 3	Large-scale confirmatory trials	Strongest
Phase 2	Exploratory trials	Moderate
Phase 1	Safety trials	Weaker

4. Literature Evidence

Literature Type	Description	Evidence Strength
RCT	Randomized Controlled Trial	Highest
Cohort	Cohort Study	Moderate
Case Report	Individual Case Report	Reference

Step-by-Step Guide

Step 1: Start from the High Evidence page
Step 2: Click on a drug of interest to enter the report
Step 3: First read the "One-Sentence Summary" and "Quick Overview Table"
Step 4: For deeper understanding, read clinical trials and literature evidence
Step 5: Check "Conclusion & Next Steps" for recommended actions

Frequently Asked Questions

Q: High prediction score but low evidence level - how to interpret?

Prediction score reflects AI confidence. Evidence level reflects clinical research adequacy. The two can differ.

For example: A drug might have a 99.97% prediction score but only L2 evidence. This means:

AI is very confident in the prediction
But clinical evidence is not yet sufficient (only early-stage research)
This is a promising research direction worth attention

Q: What’s the difference between Go and Proceed?

Decision	Meaning	Recommended Action
Go	Very strong evidence	Can proceed to in-depth evaluation
Proceed	Adequate evidence	Worth further feasibility evaluation
Consider	Some evidence	Weigh risks before deciding
Explore	Worth exploring	Gather more data first
Hold	Insufficient evidence	Not recommended to proceed currently