HbA1c Reduction on Type 2 diabetes simulation Study A (test)
92TTG (%)Capability-weighted offline RL
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Capability-weighted offline RLReward Structure=terminal reward2026.05 | 92 | 42 | 1.21 | |
| Behavior baselineConfiguration=Behavior baseline2026.05 | 91 | 27 | 1.02 | |
| Capability-weighted offline RLReward Structure=tiered reward2026.05 | 74 | 17 | 0.69 | |
| Uniform-weighted offline RLReward Structure=tiered reward2026.05 | 72 | 16 | 0.65 |