Research map / Large language models / Agents, code and math
When the Domain Expert Has No Time and the LLM Developer Has No Clinical Expertise: Real-World Lessons from LLM Co-Design in a Safety-Net Hospital
Avni Kothari, Patrick Vossler, Jean Digitale, Mohammad Forouzannia, Elise Rosenberg, Michele Lee, Jenneé Bryant, Melanie Molina, James Marks, Lucas Zier, Jean Feng
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 tiered prompt refinement (onboarding docs, CFE feedback, ICL quotes) over the base attribute prompt (ablation) | GPT-4o social needs extraction · LLM-judge concordance, overall | concordance: 0.77 -> 0.95 | Improved |
| add SW onboarding materials to attribute prompts (ablation) | Prompt1 · LLM-judge concordance, overall | concordance: 0.77 -> 0.88 | Improved |
| add cross-functional expert feedback to prompts (ablation) | Prompt2 · LLM-judge concordance, overall | concordance: 0.88 -> 0.91 | Improved |
| add ICL examples from previous SW notes and exclusion examples (ablation) | Prompt3 · LLM-judge concordance, overall | concordance: 0.91 -> 0.95 | Improved |
| add CFE feedback to Outpatient Therapy prompt (ablation) | Prompt2 · Outpatient Therapy concordance | concordance: 0.72 -> 0.65 | Hurt |
| add CFE feedback to Safety prompt (ablation) | Prompt2 · Safety concordance | concordance: 0.92 -> 0.84 | Hurt |
| tiered prompt refinement on Food Insecurity attribute (ablation) | Prompt1 · Food Insecurity concordance | concordance: 1.00 -> 0.99 | No clear effect |
| tiered prompt refinement on Reason for Consult attribute (ablation) | Prompt1 · Reason for Consult concordance | concordance: 0.54 -> 0.92 | Improved |
| tiered prompt refinement on Outpatient Therapy attribute (ablation) | Prompt1 · Outpatient Therapy concordance | concordance: 0.51 -> 0.90 | Improved |
| use LLM-as-a-judge as proxy for human validation of prompt clarity | Reason for Consult and Patient Contacts · Human validation | agreement with annotators: LLM app 0.65 (Reason for Consult), 0.72 (Patient Contacts) | Mixed |
| decompose free-form summary into 15 structured attributes | pre-chart summarization · Human validation overall | agreement with annotators: 0.89 LLM app vs 0.89 inter-annotator | Improved |
Closest papers by meaning
- MedJourney: Benchmark and Evaluation of Large Language Models over Patient Clinical Journey (NeurIPS 2024 · similarity 0.87)
- Position: Beyond Assistance – Reimagining LLMs as Ethical and Adaptive Co-Creators in Mental Health Care (ICML 2025 · similarity 0.87)
- Multilingual Medical Language Models: A Path to Improving Lay Health Worker Effectiveness (Student Abstract) (AAAI 2024 · similarity 0.86)
- Efficiently Democratizing Medical LLMs for 50 Languages via a Mixture of Language Family Experts (ICLR 2025 · similarity 0.85)
- Evaluating LLMs Across Multi-Cognitive Levels: From Medical Knowledge Mastery to Scenario-Based Problem Solving (ICML 2025 · similarity 0.85)
- Cross-Care: Assessing the Healthcare Implications of Pre-training Data on Language Model Bias (NeurIPS 2024 · similarity 0.85)
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.
