Assessing Patient Perceptions of Artificial Intelligence versus Human Surgeon Counseling in Facial Plastic Surgery
نویسندگان
1 Texas Center for Facial Plastic Surgery, Department of Otorhinolaryngology – Head and Neck Surgery, University of Texas Health Science Center Houston McGovern Medical School, Houston, TX, USA.
2 Texas Center for Facial Plastic Surgery, Department of Otorhinolaryngology – Head and Neck Surgery, University of Texas Health Science Center Houston McGovern Medical School, Houston, TX, USA.
3 Texas Center for Facial Plastic Surgery, Department of Otorhinolaryngology – Head and Neck Surgery, University of Texas Health Science Center Houston McGovern Medical School, Houston, TX, USA.
4 Texas Center for Facial Plastic Surgery, Department of Otorhinolaryngology – Head and Neck Surgery, University of Texas Health Science Center Houston McGovern Medical School, Houston, TX, USA.
5 Texas Center for Facial Plastic Surgery, Department of Otorhinolaryngology – Head and Neck Surgery, University of Texas Health Science Center Houston McGovern Medical School, Houston, TX, USA.
doi
10.22037/orlfps.v10i1.47066چکیده
Background: The integration of artificial intelligence (AI), particularly generative large language (LLM) models like ChatGPT, promises to enhance patient education and communication in many fields, including facial plastic surgery. Aim: We herein assess patient perceptions of AI-generated versus human surgeon preoperative and postoperative counseling for septorhinoplasty surgery. Methods: Responses to hypothetical questions by ChatGPT and a human surgeon were evaluated by 103 blind evaluators from a general audience to assess empathy, accuracy, completeness, and overall quality. Evaluators were also asked to quantify their prior usage of AI LLMs and rate their confidence in discerning the AI and human responses. Results: ChatGPT's responses were generally preferred, receiving significantly higher scores in accuracy (p<0.001), completeness (p<0.001), and overall quality (p<0.001). ChatGPT scored lower in empathy, but the difference was not significant (p=0.11). No significant differences were found in evaluators' trust in AI or their ability to discern between human and AI responses based on past AI LLM usage (p=0.43; p=0.25).. Conclusion: AI, specifically ChatGPT, can supplement traditional patient education and communication methods in facial plastic surgery. Further research is necessary to understand the broader implications and optimal integration of AI in clinical practice.
