MIMIC-III
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
MIMIC-III target v1.4
16.95AUPRC
10
MIMIC-III Full v1.4 (test)
0.123Macro F1
10
MIMIC-III Hypertension
47.27AUPRC
10
MIMIC-III 75%-3 task (test)
0.396MSE
9
MIMIC-III 50%-50% task (test)
0.394Inference Time (s)
9
MIMIC-III (test)
35.5AUPRC
8
MIMIC-III unseen in WS n=174 (test)
94.1Precision
8
MIMIC-III continual protocol A (time-window shift by admission year)
0.86Full-data UB Score
7
MIMIC-III Radiology n=53+/198 (External val)
89.4Precision
7
MIMIC-III sepsis MNAR 40% (test)
9.55V_hat(pi)
6
MIMIC-III sepsis 40% MNAR 20% (test)
10,408.2V_hat(pi)
6
MIMIC-III Cardiovascular African
4.71RMSE ($ au=1$)
6
MIMIC-III African subgroup OOD infectious and inflammatory diseases (out of distribution)
4.92RMSE (τ=1)
6
Semi-synthetic MIMIC-III Expanded Time-varying Confounding (test)
0.53Bias (CF 1)
5
MIMIC-III v1.4 (test)
91.58Accuracy (140-239)
5
MIMIC-III v1.4 (test)
78.1Accuracy
5
MIMIC-III-ext-VeriFact-BHC 120 prompts (test)
10.7Local Verifier NS-rate
4
MIMIC-III n=400 (val)
90.1Precision
4
MIMIC-III 1.4 (test)
87.1Onset AUROC
4
MIMIC-III ICD-9 20 tasks (test)
71.23Average AUC
4
MIMIC-III All patients
49.96Acute Renal Failure Score
3
MIMIC-III v1.4 (1000 patients)
13.5AUPRC
3
MIMIC-III discharge summaries (val)
67.6Precision
3
MIMIC-III seen in WS n=499 (test)
83.3Precision
3
MIMIC-III MIMIC-Extract pipeline (test)
89.7AUROC
3