Research map / Label-efficient and robust learning / Recognition and long-tail
MiME: Multilevel Medical Embedding of Electronic Health Records for Predictive Healthcare
Edward Choi, Cao Xiao, Walter Stewart, Jimeng Sun
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Add auxiliary diagnosis and treatment prediction tasks to MiME (ablation) | MiME · HF prediction E1 | PR-AUC: 0.2711 -> 0.2831 | Improved |
| Add auxiliary diagnosis and treatment prediction tasks to MiME (ablation) | MiME · HF prediction D3 | PR-AUC: 0.4685 -> 0.4787 | Improved |
| Model diagnosis-treatment relationships with multilevel embeddings | HF prediction · E1-E4 | PR-AUC: 0.2462 -> 0.2831 | Improved |
| Use MiME instead of the strongest baseline | Sequential disease prediction · Sequential disease prediction | Recall@20: 0.7926 -> 0.7945 | Improved |
| Use MiME instead of Med2Vec | HF prediction · D1 (visit complexity 0-15%) | PR-AUC: 0.2771 -> 0.2750 | No clear effect |
| Use tanh rather than ReLU for the visit embedding activation | GRU baseline embedding · HF prediction D3 | PR-AUC: 0.4100 -> 0.4415 | Improved |
| Add a second sigmoid MLP layer to the visit embedding | GRU baseline embedding · Sequential disease prediction | Recall@5: 0.5110 -> 0.2132 | Hurt |
| Use a sigmoid MLP visit embedding instead of a single sigmoid layer | GRU baseline embedding · HF prediction E1 | PR-AUC: 0.2354 -> 0.0843 | Hurt |
| Use Med2Vec rather than MiME for rare-disease prediction | Sequential disease prediction · Rare diagnoses (0.01%-0.05% prevalence) | Accuracy@5: 0.0965 -> 0.1156 | Improved |
| Use MiME rather than tanh_mlp | Sequential disease prediction · Diagnoses with 0.05%-0.2% prevalence | Accuracy@5: 0.2353 -> 0.2625 | Improved |
Later papers that compared with this method
2 later accepted papers report MiME as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- NeuralCohort: Cohort-aware Neural Representation Learning for Healthcare Analytics (ICML 2025 · better on 6 of 6 shared results)
- More Reliable Pseudo-labels, Better Performance: A Generalized Approach to Single Positive Multi-label Learning (ICCV 2025 · better on 4 of 4 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- BoxLM: Unifying Structures and Semantics of Medical Concepts for Diagnosis Prediction in Healthcare (ICML 2025 · similarity 0.85)
- DrFuse: Learning Disentangled Representation for Clinical Multi-Modal Fusion with Missing Modality and Modal Inconsistency (AAAI 2024 · similarity 0.84)
- DearLLM: Enhancing Personalized Healthcare via Large Language Models-Deduced Feature Correlations (AAAI 2025 · similarity 0.84)
- Risk Horizons: Structured Hypothesis Spaces for Longitudinal Clinical Prediction (NeurIPS 2026 · similarity 0.84)
- One Loss to Rule Them All: Marked Time-to-Event for Structured EHR Foundation Models (NeurIPS 2026 · similarity 0.84)
- Medformer: A Multi-Granularity Patching Transformer for Medical Time-Series Classification (NeurIPS 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
