Agent Verification on MIMIC-IV Diverticulitis
0.9975AUROCGLEAN + Active (εu = 0.0)
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| GLEAN + Active (εu = 0.0)Agent Backbone=Qwen3-30B-A3B-Instruct2026.03 | 0.9975 | 0.0309 | 0.111 | 0.0393 | |
| GLEAN + Active (εu = 0.0)Agent Backbone=Qwen2.5-7B-Instruct2026.03 | 0.9869 | 0.1049 | 0.0673 | 0.0522 | |
| GLEAN + Active (εu = 0.5)Agent Backbone=Qwen3-30B-A3B-Instruct2026.03 | 0.9862 | 0.037 | 0.1031 | 0.0453 | |
| GLEAN (K = 3)Agent Backbone=Qwen3-30B-A3B-Instruct2026.03 | 0.9794 | 0.0741 | 0.1051 | 0.0632 | |
| GLEAN + Active (εu = 0.5)Agent Backbone=Qwen2.5-7B-Instruct2026.03 | 0.9756 | 0.1049 | 0.0686 | 0.0586 | |
| GLEAN (K = 3)Agent Backbone=Qwen2.5-7B-Instruct2026.03 | 0.9568 | 0.1667 | 0.0724 | 0.0892 | |
| GLEAN (K = 1)Agent Backbone=Qwen3-30B-A3B-Instruct2026.03 | 0.9147 | 0.179 | 0.1077 | 0.1263 | |
| Self-ConsistencyAgent Backbone=Qwen3-30B-A3B-Instruct2026.03 | 0.9131 | 0.1667 | 0.1889 | 0.1608 | |
| Self-ConsistencyAgent Backbone=Qwen2.5-7B-Instruct2026.03 | 0.9011 | 0.216 | 0.188 | 0.1677 | |
| GLEAN (K = 1)Agent Backbone=Qwen2.5-7B-Instruct2026.03 | 0.8919 | 0.2469 | 0.0502 | 0.1359 | |
| RAG-AugmentedAgent Backbone=Qwen3-30B-A3B-Instruct2026.03 | 0.8577 | 0.2222 | 0.1816 | 0.1916 | |
| Semantic EntropyAgent Backbone=Qwen2.5-7B-Instruct2026.03 | 0.8573 | 0.2531 | — | — | |
| RAG-AugmentedAgent Backbone=Qwen2.5-7B-Instruct2026.03 | 0.8358 | 0.2284 | 0.2304 | 0.2271 | |
| ORMAgent Backbone=Qwen3-30B-A3B-Instruct2026.03 | 0.8228 | 0.2614 | 0.1939 | 0.2381 | |
| LLM-as-a-JudgeAgent Backbone=Qwen3-30B-A3B-Instruct2026.03 | 0.8166 | 0.2654 | 0.1172 | 0.191 | |
| Semantic EntropyAgent Backbone=Qwen3-30B-A3B-Instruct2026.03 | 0.8128 | 0.2346 | — | — | |
| Med-PRMAgent Backbone=Qwen3-30B-A3B-Instruct2026.03 | 0.8099 | 0.2531 | 0.4815 | 0.4684 | |
| Self-VerificationAgent Backbone=Qwen2.5-7B-Instruct2026.03 | 0.7891 | 0.3148 | 0.1198 | 0.2012 | |
| P(TRUE)Agent Backbone=Qwen3-30B-A3B-Instruct2026.03 | 0.7851 | 0.284 | 0.2392 | 0.2488 | |
| ORMAgent Backbone=Qwen2.5-7B-Instruct2026.03 | 0.7793 | 0.3203 | 0.1594 | 0.2168 | |
| LLM-as-a-JudgeAgent Backbone=Qwen2.5-7B-Instruct2026.03 | 0.7709 | 0.3457 | 0.2558 | 0.2859 | |
| Self-VerificationAgent Backbone=Qwen3-30B-A3B-Instruct2026.03 | 0.7681 | 0.3333 | 0.2416 | 0.2655 | |
| P(TRUE)Agent Backbone=Qwen2.5-7B-Instruct2026.03 | 0.728 | 0.3642 | 0.3445 | 0.3379 | |
| Med-PRMAgent Backbone=Qwen2.5-7B-Instruct2026.03 | 0.7111 | 0.3704 | 0.4851 | 0.4703 |