Policy Improvement on OhioT1DM
6.5Patient 596 ResultBi-Level Offline Policy Optimization Algorithm
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Bi-Level Offline Policy Optimization Algorithmexperiments=50 repeated, evaluation_protocol=Monte Carlo discounted returns2023.10 | 6.5 | 33.1 | 36.9 | 7.9 | 13.2 | 20.4 | |
| ATACexperiments=50 repeated, evaluation_protocol=Monte Carlo discounted returns2023.10 | 5.1 | 26.9 | 29.7 | 7.2 | 10.1 | 18.2 | |
| TD3+BCexperiments=50 repeated, evaluation_protocol=Monte Carlo discounted returns2023.10 | 4.8 | 22.4 | 27.8 | 7.4 | 9.7 | 17.1 | |
| OptiDICEexperiments=50 repeated, evaluation_protocol=Monte Carlo discounted returns2023.10 | 4.7 | 27.7 | 28.8 | 8.1 | 10.3 | 17.9 | |
| CQLexperiments=50 repeated, evaluation_protocol=Monte Carlo discounted returns2023.10 | 4.6 | 21.6 | 26.5 | 6.7 | 8.7 | 16.5 | |
| COMBOexperiments=50 repeated, evaluation_protocol=Monte Carlo discounted returns2023.10 | 4.1 | 27 | 30.6 | 6.8 | 9.8 | 17.5 | |
| BCQexperiments=50 repeated, evaluation_protocol=Monte Carlo discounted returns2023.10 | 3.8 | 20.3 | 24.3 | 5.7 | 7.5 | 14.3 | |
| IQLexperiments=50 repeated, evaluation_protocol=Monte Carlo discounted returns2023.10 | 3.4 | 22.7 | 25.8 | 6.1 | 7.8 | 14 | |
| BEARexperiments=50 repeated, evaluation_protocol=Monte Carlo discounted returns2023.10 | 2.7 | 22.9 | 25.6 | 5 | 5.9 | 12.7 |