Evaluating Large Language Models for Preoperative Patient Education in Superior Capsular Reconstruction: Comparative Study of Claude, GPT, and Gemini
Bibliographic record
Abstract
Background: Large language models (LLMs) are revolutionizing natural language processing, increasingly applied in clinical settings to enhance preoperative patient education. Objective: This study aimed to evaluate the effectiveness and applicability of various LLMs in preoperative patient education by analyzing their responses to superior capsular reconstruction (SCR)-related inquiries. Methods: In total, 10 sports medicine clinical experts formulated 11 SCR issues and developed preoperative patient education strategies during a webinar, inputting 12 text commands into Claude-3-Opus (Anthropic), GPT-4-Turbo (OpenAI), and Gemini-1.5-Pro (Google DeepMind). A total of 3 experts assessed the language models' responses for correctness, completeness, logic, potential harm, and overall satisfaction, while preoperative education documents were evaluated using DISCERN questionnaire and Patient Education Materials Assessment Tool instruments, and reviewed by 5 postoperative patients for readability and educational value; readability of all responses was also analyzed using the cntext package and py-readability-metrics. Results: Between July 1 and August 17, 2024, sports medicine experts and patients evaluated 33 responses and 3 preoperative patient education documents generated by 3 language models regarding SCR surgery. For the 11 query responses, clinicians rated Gemini significantly higher than Claude in all categories (P<.05) and higher than GPT in completeness, risk avoidance, and overall rating (P<.05). For the 3 educational documents, Gemini's Patient Education Materials Assessment Tool score significantly exceeded Claude's (P=.03), and patients rated Gemini's materials superior in all aspects, with significant differences in educational quality versus Claude (P=.02) and overall satisfaction versus both Claude (P<.01) and GPT (P=.01). GPT had significantly higher readability than Claude on 3 R-based metrics (P<.01). Interrater agreement was high among clinicians and fair among patients. Conclusions: Claude-3-Opus, GPT-4-Turbo, and Gemini-1.5-Pro effectively generated readable presurgical education materials but lacked citations and failed to discuss alternative treatments or the risks of forgoing SCR surgery, highlighting the need for expert oversight when using these LLMs in patient education.
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 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 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".