Research map / Large language models / Retrieval and knowledge
Retrieve, Reason, and Refine: Generating Accurate and Faithful Patient Instructions
Fenglin Liu, Bang Yang, Chenyu You, Xian Wu, Shen Ge, Zhangdaihong Liu, Xu Sun, Yang Yang, David A. Clifton
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 Re3Writer (retrieve, reason, refine) | Transformer · Patient Instruction (PI) test | BLEU-4: 27.3 -> 30.5 | Improved |
| add Re3Writer | LSTM · Patient Instruction (PI) test | BLEU-4: 21.0 -> 25.3 | Improved |
| add Re3Writer | Seq2Seq · Patient Instruction (PI) test | BLEU-4: 25.1 -> 26.8 | Improved |
| add Re3Writer | CopyNet · Patient Instruction (PI) test | BLEU-4: 24.4 -> 26.0 | Improved |
| retrieve similar diagnosis PIs only (ablation) | Transformer · Patient Instruction (PI) test | BLEU-4: 27.3 -> 28.0 | Improved |
| retrieve similar medication PIs only (ablation) | Transformer · Patient Instruction (PI) test | BLEU-4: 27.3 -> 28.9 | Improved |
| retrieve similar procedure PIs only (ablation) | Transformer · Patient Instruction (PI) test | BLEU-4: 27.3 -> 28.4 | Improved |
| combine diagnosis, medication and procedure retrieval (ablation) | Seq2Seq · Patient Instruction (PI) test | BLEU-4: 25.1 -> 26.3 | Improved |
| combine all three retrieval types (ablation) | Transformer · Patient Instruction (PI) test | BLEU-4: 27.3 -> 29.2 | Improved |
| add Reason knowledge graph on top of retrieval (ablation) | Transformer · Patient Instruction (PI) test | BLEU-4: 29.2 -> 29.9 | Improved |
| add Refine gating on top of retrieve and reason (ablation) | Transformer · Patient Instruction (PI) test | BLEU-4: 29.9 -> 30.5 | Improved |
| add Refine gating (ablation) | Seq2Seq · Patient Instruction (PI) test | BLEU-4: 26.6 -> 26.8 | Improved |
Closest papers by meaning
- MedAlign: A Clinician-Generated Dataset for Instruction Following with Electronic Medical Records (AAAI 2024 · similarity 0.83)
- CLR-voyance : Reinforcing Open-Ended Reasoning for Inpatient Clinical Decision Support with Outcome-Aware Rubrics (NeurIPS 2026 · similarity 0.83)
- PatientSim: A Persona-Driven Simulator for Realistic Doctor-Patient Interactions (NeurIPS 2025 · similarity 0.82)
- MedEvoEval: Evaluating Continual Evolution of Doctor Agents through Simulated Clinical Episodes (NeurIPS 2026 · similarity 0.82)
- Position: Retrieval-augmented systems can be dangerous medical communicators (ICML 2025 · similarity 0.81)
- PhysicianBench: Evaluating LLM Agents in Real-World EHR Environments (NeurIPS 2026 · similarity 0.81)
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.
