List your company

EHR Multi-turn Reasoning on MIMIC III (OOD)

0.0395DiagnosesReAct
0.0376040.0504020.06320.075998Jan 20, 2026

Evaluation results

MethodLinks
ReActEvolved=false, Backbone=Qwen3-30B-A3B2026.010.03950.14180.18710.08410.01160.32720.1319
ReflexionEvolved=false, Backbone=Qwen3-30B-A3B2026.010.04170.1340.1710.07150.01330.26560.1162
ReflecTool (IR)Evolved=true, Backbone=Qwen3-30B-A3B2026.010.04680.18750.16020.0780.02820.35760.1431
ReSumEvolved=false, Backbone=Qwen3-30B-A3B2026.010.05020.14170.17640.08510.01850.24850.1201
ReflecTool (CS)Evolved=true, Backbone=Qwen3-30B-A3B2026.010.05860.19560.15950.07820.02980.36890.1484
ReasoningBankEvolved=true, Backbone=Qwen3-30B-A3B2026.010.06020.20870.17360.08150.0170.17220.1189
RETROSUMEvolved=false, Backbone=Qwen3-30B-A3B2026.010.07050.1970.1510.08910.03370.36270.1507
RETROSUMEvolved=true, Backbone=Qwen3-30B-A3B2026.010.08690.1860.16910.08070.04730.3570.1545