Scholarlink

Aim <p>This study aimed to assess the reliability of orthopaedic information provided by ChatGPT in response to common patient inquiries and concerns related to Total Joint Replacement surgery, focusing on preventing the dissemination of potentially harmful medical advice.</p> Method <p>This qualitative exploratory case study was conducted at a tertiary care centre hospital. Ten common questions patients pose to orthopaedic surgeons when considering knee arthroplasty were formulated and presented independently to an Orthopaedic Consultant, Associate Consultant, a Fellow in Joint Replacement surgery, and ChatGPT. For review, the answers were submitted to a panel of three orthopaedic surgeons specializing in arthroplasty. They were scored based on accuracy, relevance, and usefulness, with a maximum possible score of 100.0 points, allowing for 0.5-point increments. They were also asked to identify which answers were from chat GPT, not humans. </p> Results <p>ChatGPT exhibited the highest total aggregate score of 232.0 points out of a maximum of 300.0 points, surpassing the scores of the human participants (Participant 1: 197.0 points; Participant 2: 227.5 points; Participant 3: 220.5 points). Furthermore, two out of three panel specialists rated ChatGPT the highest. When comparing the average scores for ChatGPT and the human participants for each question, ChatGPT outperformed the human participants in 8 out of 10 questions. Out of the 120 encounter instances, the evaluator could only point correctly that the response was from ChatGPT response 14 times (11.66%).</p> Conclusion <p>This study highlights the utility and limitations of ChatGPT in the medical field—ChatGPT exhibits great potential in assisting doctors and surgeons in patient care by providing accurate and relevant information. The study also demonstrated that the answers seemed indistinguishable from humans in most cases. In the current landscape of ChatGPT and other AI technologies, their integration in the medical field should be viewed as complementary to human expertise, which must be leveraged for the greater good. </p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Is ChatGPT more Humane than Humans?—Accuracy, Relevance and humanness of the Answers Provided by ChatGPT for Patient Education for Total Knee Replacement

  • 期刊论文

  • 发表日期：

    2025年7月3日

查看全文
  • Khushi Bafna,
  • Amit Kumar Yadav,
  • Shaurya Bagaria,
  • Ayushi Gianchandani,
  • Murali Poduval,
  • Vaibhav Bagaria

摘要

Aim

This study aimed to assess the reliability of orthopaedic information provided by ChatGPT in response to common patient inquiries and concerns related to Total Joint Replacement surgery, focusing on preventing the dissemination of potentially harmful medical advice.

Method

This qualitative exploratory case study was conducted at a tertiary care centre hospital. Ten common questions patients pose to orthopaedic surgeons when considering knee arthroplasty were formulated and presented independently to an Orthopaedic Consultant, Associate Consultant, a Fellow in Joint Replacement surgery, and ChatGPT. For review, the answers were submitted to a panel of three orthopaedic surgeons specializing in arthroplasty. They were scored based on accuracy, relevance, and usefulness, with a maximum possible score of 100.0 points, allowing for 0.5-point increments. They were also asked to identify which answers were from chat GPT, not humans.

Results

ChatGPT exhibited the highest total aggregate score of 232.0 points out of a maximum of 300.0 points, surpassing the scores of the human participants (Participant 1: 197.0 points; Participant 2: 227.5 points; Participant 3: 220.5 points). Furthermore, two out of three panel specialists rated ChatGPT the highest. When comparing the average scores for ChatGPT and the human participants for each question, ChatGPT outperformed the human participants in 8 out of 10 questions. Out of the 120 encounter instances, the evaluator could only point correctly that the response was from ChatGPT response 14 times (11.66%).

Conclusion

This study highlights the utility and limitations of ChatGPT in the medical field—ChatGPT exhibits great potential in assisting doctors and surgeons in patient care by providing accurate and relevant information. The study also demonstrated that the answers seemed indistinguishable from humans in most cases. In the current landscape of ChatGPT and other AI technologies, their integration in the medical field should be viewed as complementary to human expertise, which must be leveraged for the greater good.