Key Takeaways
- ChatGPT-4 generated largely accurate and appropriate responses to six common postoperative concerns after lower lid blepharoplasty, with mean accuracy scores above 4.3 out of 5.
- Most oculoplastic surgeons reviewing the responses (75%–100%) judged the management recommendations to be appropriate, depending on the complication addressed.
- While responses were highly understandable, readability levels were relatively high and consistent physician oversight was still needed before clinical reliance.
ChatGPT-4 produced generally accurate, appropriate, and understandable responses to common postoperative concerns after lower lid blepharoplasty, but it still requires further refinement and physician oversight before it can be relied on as a clinical resource, according to a study.
Using a cross-sectional design, the study examined ChatGPT-4’s answers to 6 postoperative issues:
- Bleeding
- Diplopia
- Dry eyes
- Pain
- Lower lid retraction
- Bruising, hematoma, or orbital hemorrhage
Twelve physicians from the American Society of Ophthalmic Plastic and Reconstructive Surgery graded the responses for clinical accuracy using a 5-point Likert scale and judged whether the management recommendations were appropriate.
Average accuracy scores across the 6 responses ranged from 4.33 to 4.67 out of 5, with an overall mean score of 4.50 ± 0.63. Between 75% and 100% of graders considered the management recommendations appropriate, depending on the topic.
Readability analysis showed Flesch Reading Ease scores between 28.9 and 50.2, corresponding to Flesch-Kincaid Grade Levels from 9.4 to 13.4. The average understandability score was 91.67 ± 0.52%, and all responses received an actionability score of 60%.
Reference
Hofer LK, Hofer LK, Nakasian NW, et al. ChatGPT as a Postoperative Tool Following Lower Lid Blepharoplasty. Ophthalmic Plast Reconstr Surg. 2025;doi: 10.1097/IOP.0000000000003156. Epub ahead of print. PMID: 41474115.
Related Content
