Clinical Patient State Learning on MIMIC-III Appendix benchmark
67.9FQEOPL-MT-MNAR
Evaluation results
| Method | Links | ||
|---|---|---|---|
| OPL-MT-MNARMissingness Handling=Explicit MNAR + DocProcess + Text2026.04 | 67.9 | 88.6 | |
| MedDreamerMissingness Handling=World model + AFI2026.04 | 58.3 | 86.7 | |
| mTANDMissingness Handling=Multi-time attention2026.04 | 52.3 | 85.8 | |
| BRITSMissingness Handling=Bidirectional imputation2026.04 | 51.6 | 85.2 | |
| GRU-DMissingness Handling=Implicit time decay2026.04 | 50.8 | 84.4 | |
| Forward Fill + LSTMMissingness Handling=Sample-and-hold2026.04 | 48.8 | 83.8 | |
| Mean Imputation + LSTMMissingness Handling=Impute to mean2026.04 | 48.3 | 83.3 |