ChatGPT v.s. Human in Writing a Preoperative Visit Sheet
Study Details
Study Description
Brief Summary
The goal of this study is to evaluate the performance of large language models i.e. ChatGPT, in making preoperative visit sheets using clinical records. The main questions it aims to answer are:
-
Can large language models read clinical records and make preoperative visit sheets same as physicians?
-
Can physicians distinguish preoperative visit sheets made by physicians and models?
Participants' records will be generated using ChatGPT-4, and read by both ChatGPT-4 and physicians to make 2 separate preoperative visit sheets, and form 2 groups, GPT group and physician group respectively.
A group of professionals will compare result of above mentioned 2 groups to see if ChatGPT can afford to write preoperative visit sheets.
Condition or Disease | Intervention/Treatment | Phase |
---|---|---|
|
Study Design
Arms and Interventions
Arm | Intervention/Treatment |
---|---|
Experiment
|
Other: ChatGPT-4
Subjects' clinical records will be read by ChatGPT-4 to write preoperative visit sheets.
|
Control
|
Other: Clinician
Subjects' clinical records will be read by clinicians to write preoperative visit sheets.
|
Outcome Measures
Primary Outcome Measures
- Familiarity to human writing [Each record will be screened for 6 minutes at most and results will be given in 8 minutes after receiving a record.]
Based on professionals' subjective evaluation, familiarity is a binary result indicating weather professionals think the preoperative visit sheet was written by physicians or ChatGPT, where 0 indicates "not written by human" and 1 indicates "written by human".
- Satisfaction for clinical use [Each record will be screened for 6 minutes at most and results will be given in 8 minutes after receiving a record.]
A grading score of 1-10 indicating weather the professionals think the preoperative visit sheet is suitable in clinical use.
Eligibility Criteria
Criteria
Inclusion Criteria:
- Any as records are generated using ChatGPT-4.
Exclusion Criteria:
- Any as records are generated using ChatGPT-4.
Contacts and Locations
Locations
No locations specified.Sponsors and Collaborators
- Boston Intelligent Medical Research Center, Shenzhen United Scheme Technology Co., Ltd.
- Tsinghua University
Investigators
None specified.Study Documents (Full-Text)
None provided.More Information
Publications
None provided.- US2023002