Predicting Postoperative Outcomes in Full-Thickness Macular Hole Repair Surgery: ChatGPT vs. Clinical Decision.

Key Points

• This research aims to evaluate the accuracy of ChatGPT-5 in predicting outcomes of full-thickness macular hole repairs compared to traditional clinical judgment.
• Comparison of ChatGPT-5 predictions with clinical assessments
• Analysis of closure rates and visual recovery estimates
• Identification of biases in AI predictions
• ChatGPT-5 showed a high degree of predicted accuracy for surgery outcomes
• Predictions were influenced by an optimism bias, leading to overestimated results
• Clinical judgment was found to be more reliable for actual outcomes