AI Prediction Only Drugs
TxGNN model predictions without clinical evidence - research hypotheses for exploration
Evidence Standards
| Level | Definition | Clinical Significance |
|---|---|---|
| L5 | AI prediction only, no clinical evidence | Research hypothesis, requires validation |
Caution: L5 predictions are based on knowledge graph analysis only. They should NOT be used for clinical decision-making and require rigorous validation before any therapeutic application.
Statistics
| Metric | Value |
|---|---|
| Total L5 Predictions | 32,368 |
| Unique Drugs | 638 |
| Unique Indications | 4,570 |
Highest Confidence Predictions
These L5 predictions have the highest TxGNN scores:
| Drug | Indication | Score |
|---|---|---|
| TRAVOPROST | visceral calciphylaxis | 0.9999 |
| CAPLACIZUMAB | primary release disorder of platelets | 0.9999 |
| ROMIPLOSTIM | primary release disorder of platelets | 0.9999 |
| CAPLACIZUMAB | pseudo-von Willebrand disease | 0.9999 |
| AZATHIOPRINE | rheumatoid arthritis | 0.9999 |
| TRAVOPROST | arterial thoracic outlet syndrome | 0.9999 |
| CAPLACIZUMAB | Glanzmann thrombasthenia | 0.9999 |
| INFLIXIMAB | Crohn disease | 0.9999 |
| ADALIMUMAB | psoriatic arthritis | 0.9999 |
| USTEKINUMAB | inflammatory bowel disease | 0.9999 |
Research Applications
L5 predictions can be useful for:
- Hypothesis Generation: Starting points for research
- Literature Review: Prioritize which associations to investigate
- Grant Writing: Preliminary data for research proposals
- Drug Discovery: Identify candidates for preclinical testing
Validation Pathway
To upgrade from L5 to higher evidence levels:
L5 (AI Prediction)
↓ Literature search
L4 (Preclinical/Mechanistic)
↓ Observational studies
L3 (Cohort/Case-control)
↓ Phase 2 trials
L2 (Single RCT)
↓ Multiple Phase 3 trials
L1 (Strong Evidence)
Download L5 Data
All L5 predictions available at: