Table 1. Percentage of correct answers for each LLM in subgroups (%).
Data are expressed as the mean±95% CI. Significant differences are indicated using asterisks or tagger.
*: p<0.05, ChatGPT-4 vs Gemini 1.0. **:p<0.05, Claude 3 Opus vs Gemini 1.0. †: p<0.05, ChatGPT-4 vs Claude 3 Opus; CI: confidence interval; LLM: large language model
| Subject area | ChatGPT-4 (95% CI) | Claude３Opus (95% CI) | Gemini 1.0 (95% CI) | Statistical significance |
| Basic physiology for systemic management | 49.49 (34.67-64.31) | 57.59 (46.77-68.4) | 33.17 (21.16-45.17) | *, ** |
| Local anesthesia | 30.93 (17.43-44.43) | 36.56 (25.52-47.59) | 23.3 (10.76-35.84) | |
| Sedation and general anesthesia | 50.4 (38.99-61.81) | 44.47 (38.84-50.1) | 25.98 (21.68-30.27) | *, ** |
| Diseases that cause systemic management problems and management methods | 62.5 (47.16-77.84) | 49.61 (43.98-55.24) | 39.13 (32.01-46.26) | *, ** |
| Pain management | 49.33 (34.48-64.19) | 61.53 (50.44-72.63) | 27.24 (14.51-39.98) | ** |
| Shock and cardiopulmonary resuscitation | 43.61 (29.36-57.87) | 24.17 (14.7-33.63) | 15.56 (-2.921-34.03) | *, † |