Research map / Vision-language and multimodal / Human and embodied modeling
Learning Conceptual-Contextual Embeddings for Medical Text
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| encode concepts with contextual LSTM representations and UMLS knowledge rather than TransE entity embeddings | UMLS entity prediction · UMLS test triplets | filtered Hits@10 (%): 16.7 -> 65.4 | Helped |
| add discriminative training with similar-name negative concepts (ablation) | CC-LSTM · UMLS test triplets | filtered Hits@10 (%): 51.9 -> 65.4 | Helped |
| replace the contextual LSTM encoder with a DNN (ablation) | CC-LSTM · UMLS test triplets | filtered Hits@10 (%): 51.9 -> 27.7 | Hurt |
| use concept-name contextual representations to score language-inferable knowledge | CC-LSTM (DT) · UMLS language-inferable examples | Hits@10 (%): 77.6 | Helped |
| rely on concept text to encode non-language-inferable facts | CC-LSTM (DT) · UMLS non-language-inferable examples | Hits@10 (%): 20.8 | Hurt |
| replace the LSTM text encoder with the pretrained CC-LSTM | readmission classifier · MIMIC-III readmission prediction | A.R.: 0.794 -> 0.804 | Helped |
| replace the LSTM text encoder with the pretrained CC-LSTM | 30-day mortality classifier · MIMIC-III 30-day post-discharge mortality | A.R.: 0.823 -> 0.839 | Helped |
| replace the LSTM text encoder with the pretrained CC-LSTM | 1-year mortality classifier · MIMIC-III 1-year post-discharge mortality | A.R.: 0.820 -> 0.837 | Helped |
| replace the ESIM encoding LSTM with the pretrained CC-LSTM | ESIM · MedNLI dev | Acc: 74.8 -> 77.1 | Helped |
| replace the ESIM encoding LSTM with the pretrained CC-LSTM | ESIM · MedNLI test | Acc: 71.3 -> 75.2 | Helped |
Closest papers by meaning
- Reasoning-Enhanced Healthcare Predictions with Knowledge Graph Community Retrieval (ICLR 2025 · similarity 0.85)
- BoxLM: Unifying Structures and Semantics of Medical Concepts for Diagnosis Prediction in Healthcare (ICML 2025 · similarity 0.85)
- SMMILE: An expert-driven benchmark for multimodal medical in-context learning (NeurIPS 2025 · similarity 0.83)
- SM3-Text-to-Query: Synthetic Multi-Model Medical Text-to-Query Benchmark (NeurIPS 2024 · similarity 0.83)
- MedAlign: A Clinician-Generated Dataset for Instruction Following with Electronic Medical Records (AAAI 2024 · similarity 0.83)
- Multimodal Medical Code Tokenizer (ICML 2025 · similarity 0.83)
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.