Table 2.
Rates of consistent differential diagnoses made by two independent physicians and those made within the ten lists by ChatGPT-3.
| Variable | ChatGPT-3 1 | Consistent Differential Diagnoses by Two Physicians | |
|---|---|---|---|
| total, n(%) | 62/88 (70.5) | 88/150 (58.7) | |
| 1. | abdominal pain, n(%) | 7/9 (77.8) | 9/15 (60.0) |
| 2. | fever, n(%) | 4/8 (50.0) | 8/15 (53.3) |
| 3. | chest pain, n(%) | 9/11 (81.8) | 11/15 (73.3) |
| 4. | breathing difficulty, n(%) | 7/8 (87.5) | 8/15 (53.3) |
| 5. | joint pain, n(%) | 7/8 (87.5) | 8/15 (53.3) |
| 6. | vomiting, n(%) | 6/9 (66.7) | 9/15 (60.0) |
| 7. | ataxia/difficulty walking, n(%) | 6/10 (60.0) | 10/15 (66.7) |
| 8. | back pain, n(%) | 5/7 (71.4) | 7/15 (46.7) |
| 9. | cough, n(%) | 4/10 (40.0) | 10/15 (66.7) |
| 10. | dizziness, n(%) | 7/8 (87.5) | 8/15 (53.3) |
1 The consistent differential diagnoses between the two physicians were used as the denominator.