Human-Centered Prompt Engineering Framework Guides Clinical Summaries
Mieke Deschepper and colleagues at Ghent University Hospital (2026) develop a human-centered framework for creating individualized prompts to guide large language models in summarizing medical discharge letters. They ran a workshop with 26 clinicians that generated 170 ideas and a 110-item questionnaire completed by 33 participants, producing CO-STAR–refined, role-specific prompts emphasizing follow-up, medical history, and structure. The pipeline is design-focused and requires empirical validation before clinical deployment.
Key Points
- 1Generates individualized prompts using CO-STAR informed by 170 workshop ideas and a 110-item survey
- 2Highlights follow-up, medical history, and structure as most valued summary components by clinicians
- 3Provides a scalable, human-centered prompt pipeline requiring empirical validation before clinical deployment
Scoring Rationale
Methodological novelty and clinical relevance drive the score, limited by absence of empirical validation and outcome data.
Sources
Public references used for this report.
Practice with real Health & Insurance data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Health & Insurance problems
