Evaluating ChatGPT-4 Plus in Ophthalmology: Effect of Image Recognition and Domain-Specific Pretraining on Diagnostic Performance
Bibliographic record
Abstract
Background/Objectives: In recent years, the rapid advancements in artificial intelligence models, such as ChatGPT (version of 29 April 2024), have prompted interest from numerous domains of medicine, such as ophthalmology. As such, research is necessary to further assess its potential while simultaneously evaluating its shortcomings. Our study thus evaluates ChatGPT-4’s performance on the American Academy of Ophthalmology’s (AAO) Basic and Clinical Science Course (BCSC) Self-Assessment Program, focusing on its image recognition capabilities and its enhancement with domain-specific pretraining. Methods: The chatbot was tested on 1300 BCSC Self-Assessment Program questions, including text and image-based questions. Domain-specific pretraining was tested for performance improvements. The primary outcome was the model’s accuracy when presented with text and image-based multiple choice questions. Logistic regression and post hoc analyzes examined performance variations by question difficulty, image presence, and subspecialties. Results: The chatbot achieved an average accuracy of 78% compared with the average test-taker score of 74%. The repeatability kappa was 0.85 (95% CI: 0.82–0.87). Following domain-specific pretraining, the model’s overall accuracy increased to 85%. The accuracy of the model’s responses first depends on question difficulty (LR = 366), followed by image presence (LR = 108) and exam section (LR = 79). Conclusions: The chatbot appeared to be similar or superior to human trainee test takers in ophthalmology, even with image recognition questions. Domain-specific training appeared to have improved accuracy. While these results do not necessarily imply that the chatbot has the comprehensive skill level of a human ophthalmologist, the results suggest there may be educational value to these tools if additional investigations provide similar results.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.006 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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 teacher head, 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".