| dc.rights.license | CC-BY-NC-ND | |
| dc.contributor.advisor | Brinkkemper, Sjaak | |
| dc.contributor.author | Mao, Xinyu | |
| dc.date.accessioned | 2025-02-01T00:01:15Z | |
| dc.date.available | 2025-02-01T00:01:15Z | |
| dc.date.issued | 2025 | |
| dc.identifier.uri | https://studenttheses.uu.nl/handle/20.500.12932/48429 | |
| dc.description.abstract | In Netherlands, doctor spend more than 40% of their workload in writing medical report which caused too much administrative burden. Currently, Caare2Report found out a sulution which is recording what the doctor and patient talking and transcrip into text. Then using GPT to formulate medical report via transcrips. This thesis focuses on improving the quality of automated medical report. Finally, this research define 14 different repair prompts and extract 8 prompt patterns from them to improve the medical report. As a result, the accurancy of the report is more than 70% which express a highly consistent of the human report. | |
| dc.description.sponsorship | Utrecht University | |
| dc.language.iso | EN | |
| dc.subject | This thesis is aimed to define repair prpompts and prompt patterns which are used in automated medical reporting. | |
| dc.title | Prompt repairs and prompt patterns: Improving prompt engineering for automated medical reporting | |
| dc.type.content | Master Thesis | |
| dc.rights.accessrights | Open Access | |
| dc.subject.courseuu | Business Informatics | |
| dc.thesis.id | 28578 | |
