Continuous-time Patient Treatment Recommendation on MIMIC-III Sepsis PINN-based simulator (test)
87.6AIRCPO-O
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| CPO-OBase Policy Optimization Algorithm=CPO, Time-adaptation Mechanism=Optimal, K=82026.05 | 87.6 | 7.85 | 6.23 | 100 | |
| TRPO-OBase Policy Optimization Algorithm=TRPO, Time-adaptation Mechanism=Optimal, K=82026.05 | 71.56 | 8.59 | 9.75 | 3.2 | |
| CPO-EBase Policy Optimization Algorithm=CPO, Time-adaptation Mechanism=Equidistant, K=82026.05 | 53.4 | 8.15 | 8.53 | 0.6 | |
| SAC-EBase Policy Optimization Algorithm=SAC, Time-adaptation Mechanism=Equidistant, K=82026.05 | 47.54 | 8.25 | 8.24 | 0 | |
| TRPO-EBase Policy Optimization Algorithm=TRPO, Time-adaptation Mechanism=Equidistant, K=82026.05 | 24.85 | 8.08 | 8.54 | 0.2 | |
| PCPO-OBase Policy Optimization Algorithm=PCPO, Time-adaptation Mechanism=Optimal, K=82026.05 | 15.99 | 7.92 | 5.83 | 99.6 | |
| PCPO-EBase Policy Optimization Algorithm=PCPO, Time-adaptation Mechanism=Equidistant, K=82026.05 | 7.44 | 8.06 | 7.85 | 4.6 | |
| SAC-OBase Policy Optimization Algorithm=SAC, Time-adaptation Mechanism=Optimal, K=82026.05 | 1.17 | 8.92 | 6.6 | 25.8 |