Understanding Evidence Levels

MyTxGNN uses a 5-level evidence classification system (L1-L5) to help researchers quickly assess the clinical validity of drug repurposing predictions. L1 represents the highest evidence quality, while L5 indicates AI prediction only.


The 5 Evidence Levels

Level Name Definition Clinical Implication
L1 Very High Multiple Phase 3 RCTs or Systematic Reviews Strong support for clinical consideration
L2 High Single RCT or Multiple Phase 2 trials Design validation studies
L3 Moderate Phase 1/2 trials or Observational studies Requires more research
L4 Low Case reports or Preclinical only Exploratory research
L5 Prediction Only AI prediction, no clinical evidence Hypothesis generation

Detailed Explanations

L1: Very High Evidence

What qualifies:
  • Multiple Phase 3 randomized controlled trials
  • Systematic reviews with meta-analysis
  • Regulatory approval for the new indication

Recommended Action: Can proceed to clinical evaluation planning

L2: High Evidence

What qualifies:
  • Single well-designed RCT
  • Multiple Phase 2 clinical trials
  • Strong observational study with large sample

Recommended Action: Worth pursuing validation studies

L3: Moderate Evidence

What qualifies:
  • Phase 1 or early Phase 2 trials
  • Cohort or case-control studies
  • Mechanistic studies with clinical endpoints

Recommended Action: Monitor for additional evidence

L4: Low Evidence

What qualifies:
  • Case reports or case series
  • Preclinical studies only (animal/in vitro)
  • Expert opinion with mechanism support

Recommended Action: Exploratory research only

L5: Prediction Only

What qualifies:
  • TxGNN AI prediction only
  • Knowledge graph relationship inference
  • No clinical human studies found

Recommended Action: Use for hypothesis generation


Evidence vs. Prediction Score

A high TxGNN prediction score does NOT guarantee high evidence level. The prediction score reflects AI confidence based on knowledge graph patterns, while evidence level reflects actual clinical validation.
Scenario Prediction Score Evidence Level Interpretation
A 99%+ L1 Excellent - AI prediction validated by strong evidence
B 99%+ L5 Novel prediction - worth investigating but unvalidated
C 70% L1 Validated relationship but AI uncertain
D 70% L5 Low priority - uncertain and unvalidated

How to Use Evidence Levels

For Researchers

  1. Prioritize L1-L2 for immediate investigation
  2. Monitor L3 for emerging evidence
  3. Use L4-L5 for hypothesis generation and grant applications

For Clinicians

  1. L1 only should be considered for clinical discussion
  2. L2-L3 may inform clinical trial enrollment decisions
  3. L4-L5 are for research context only

Further Reading


Disclaimer
Evidence levels are assigned based on available literature and may not reflect the most recent publications. Always verify with primary sources for clinical decisions.

Hak Cipta © 2026 Projek MyTxGNN. Untuk tujuan penyelidikan sahaja. Bukan nasihat perubatan.