Evaluating ChatGPT on Orbital and Oculofacial Disorders: Accuracy and Readability Insights
Bibliographic record
Abstract
PURPOSE: To assess the accuracy and readability of responses generated by the artificial intelligence model, ChatGPT (version 4.0), to questions related to 10 essential domains of orbital and oculofacial disease. METHODS: A set of 100 questions related to the diagnosis, treatment, and interpretation of orbital and oculofacial diseases was posed to ChatGPT 4.0. Responses were evaluated by a panel of 7 experts based on appropriateness and accuracy, with performance scores measured on a 7-item Likert scale. Inter-rater reliability was determined via the intraclass correlation coefficient. RESULTS: The artificial intelligence model demonstrated accurate and consistent performance across all 10 domains of orbital and oculofacial disease, with an average appropriateness score of 5.3/6.0 ("mostly appropriate" to "completely appropriate"). Domains of cavernous sinus fistula, retrobulbar hemorrhage, and blepharospasm had the highest domain scores (average scores of 5.5 to 5.6), while the proptosis domain had the lowest (average score of 5.0/6.0). The intraclass correlation coefficient was 0.64 (95% CI: 0.52 to 0.74), reflecting moderate inter-rater reliability. The responses exhibited a high reading-level complexity, representing the comprehension levels of a college or graduate education. CONCLUSIONS: This study demonstrates the potential of ChatGPT 4.0 to provide accurate information in the field of ophthalmology, specifically orbital and oculofacial disease. However, challenges remain in ensuring accurate and comprehensive responses across all disease domains. Future improvements should focus on refining the model's correctness and eventually expanding the scope to visual data interpretation. Our results highlight the vast potential for artificial intelligence in educational and clinical ophthalmology contexts.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.020 | 0.114 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.003 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".