Comparing Performance of Large Language Model-Based Tools on Patient-Driven Glaucoma Inquiries
Bibliographic record
Abstract
PRÉCIS: GPT-4o and GPT-4o Mini outperformed Gemini Pro in effectively answering glaucoma-related questions, suggesting that GPT models provide high-quality information and highlights the potential of AI chatbots to deliver medically relevant knowledge. PURPOSE: Large language models (LLMs) can assist patients who seek medical knowledge online to guide their own glaucoma care. Understanding the differences in LLM performance on glaucoma-related questions can inform patients about the best resources to obtain relevant information. METHODS: This cross-sectional study evaluated the accuracy, comprehensiveness, quality, and readability of LLM-generated responses to glaucoma inquiries. Seven questions posted by patients on the American Academy of Ophthalmology's Eye Care Forum were randomly selected and prompted into GPT-4o, GPT-4o Mini, Gemini Pro, and Gemini Flash in September 2024. Four physicians practicing ophthalmology assessed responses using a Likert scale based on accuracy, comprehensiveness, and quality. The Flesch-Kincaid Grade level measured readability, while Bidirectional Encoder Representations from Transformers (BERT) Scores measured semantic similarity between LLM responses. Statistical analysis involved either the Kruskal-Wallis test with Dunn post-hoc test or ANOVA analysis with Tukey Honestly Significant Difference (HSD) test. RESULTS: GPT-4o rated higher in accuracy ( P =0.016), comprehensiveness ( P =0.007), and quality ( P =0.002) compared with Gemini Pro. GPT-4o Mini rated higher in comprehensiveness ( P =0.011) and quality ( P =0.007). Gemini Flash and Gemini Pro were similar across all criteria. There were no differences in readability, and LLMs mostly produced semantically similar responses. CONCLUSIONS: GPT models surpass Gemini Pro in addressing commonly asked questions about glaucoma, providing valuable insights into the application of LLMs for providing health information.
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 | 0.000 | 0.000 |
| 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".