Aesthetic Surgery Advice and Counseling from Artificial Intelligence: A Rhinoplasty Consultation with ChatGPT

Overview

ChatGPT is a large language model that produces human like text. The authors ran an observational study to see whether it could give informative and accurate answers to questions simulating a first rhinoplasty consultation. The nine questions were taken from a checklist published by the American Society of Plastic Surgeons, and the responses were assessed by specialist plastic surgeons with extensive rhinoplasty experience.

Sections of note

• Design: observational study. Level of evidence V.
• Nine questions were prompted, sourced from an American Society of Plastic Surgeons patient checklist.
• Responses were judged on three criteria: accessibility, informativeness and accuracy.
• Assessors were specialist plastic surgeons with extensive rhinoplasty experience.
• Finding: the answers were coherent and easily comprehensible, showing handling of natural language in a health context.
• The responses emphasized the importance of an individualized approach in aesthetic surgery.
• Identified limitation: the model could not provide more detailed or personalized advice.
• No scoring scale, numeric ratings or inter-rater agreement figures appear in the abstract.

What it means for a patient

• A language model can produce readable general information about rhinoplasty, but it cannot examine your nose or advise on your specific case.
• The surgeons noted the answers stayed general. Detail and personalization were where the model fell short.
• The authors suggest the value lies where access to medical advice is limited or where a patient hesitates to ask a professional.
• Limits: nine questions, no scoring scale reported, no comparison with information from surgeons or from patient websites, and a small number of assessors.

Why this paper matters

Patients research procedures online before ever booking a consultation, and language models have become part of that first step. Documenting what these tools get right and where they stop short is the first stage of understanding their role. The risks of relying on them, and their accuracy across a wider range of questions, remain unassessed.

Terms

• Large language model: software trained on text that generates human like written responses.
• ChatGPT: a publicly available conversational large language model.
• Observational study: research that records what happens without assigning any intervention.
• Level of evidence V: the weakest evidence grade, covering expert opinion and observations.
• Individualized approach: tailoring surgical planning to one patient's anatomy and goals.