Table 1 Surgeon evaluation breakdown for ChatGPT-generated biopsy and surgical pathology reports based on questionnaire criteria.
ChatGPT-generated biopsy reports | |||||
|---|---|---|---|---|---|
Questions: | Strongly disagree | Somewhat disagree | Neutral | Somewhat agree | Strongly agree |
Structure: 1. The ChatGPT report is well-structured and easy to follow. | 0 (0%) | 3 (1%) | 67 (30%) | 115 (51%) | 40 (18%) |
Readability: 2. The report clearly states the diagnosis in terms that a patient could easily understand. | 1 (1%) | 7 (3%) | 43 (19%) | 77 (34%) | 97 (43%) |
Accuracy: 3. The report correctly describes the biopsy results based on the given data. | 0 (0%) | 7 (3%) | 12 (5%) | 93 (42%) | 113 (50%) |
Disease extent explanation: 4. The report explains the extent of the disease accurately. | 0 (0%) | 13 (6%) | 28 (12%) | 69 (31%) | 115 (51%) |
Gleason grade interpretation: 5. The report interprets Gleason grade accurately. | 2 (1%) | 4 (2%) | 12 (5%) | 89 (40%) | 118 (52%) |
Comparative effectiveness: 6. Compared to the original report, the ChatGPT report will improve patients’ understanding of their disease. | 0 (0%) | 6 (3%) | 53 (24%) | 120 (53%) | 46 (20%) |
Clinical utility: 7. You are willing to use patient-centered generative AI reports in your practice. | 0 (0%) | 2 (1%) | 8 (4%) | 75 (33%) | 140 (62%) |
ChatGPT-generated surgical pathology reports: | |||||
|---|---|---|---|---|---|
Questions: | Strongly disagree | Somewhat disagree | Neutral | Somewhat agree | Strongly agree |
Structure: 1. The ChatGPT report is well-structured and easy to follow. | 0 (0%) | 0 (0%) | 9 (4%) | 75 (33%) | 141 (63%) |
Readability: 2. The report correctly explains complex medical terms and procedures in patient-friendly language. | 0 (0%) | 0 (0%) | 11 (5%) | 58 (26%) | 156 (69%) |
Accuracy: 3. The report correctly describes the pathology results based on the given data. | 0 (0%) | 1 (1%) | 43 (19%) | 23 (10%) | 158 (70%) |
Disease extent explanation: 4. The report explains the extent of the disease accurately (pathological stage, extraprostatic extension, seminal vesicle invasion, and lymph node involvement). | 0 (0%) | 1 (1%) | 17 (7%) | 96 (43%) | 111 (49%) |
Gleason grade interpretation: 5. The report interprets Gleason grade accurately. | 0 (0%) | 0 (0%) | 27 (12%) | 60 (27%) | 138 (61%) |
Surgical margin explanation: 6. The report explains surgical margins accurately. | 0 (0%) | 15 (7%) | 34 (15%) | 41 (18%) | 135 (60%) |
Comparative effectiveness: 7. Compared to the original report, the ChatGPT report will improve patients’ understanding of their disease. | 0 (0%) | 1 (1%) | 4 (2%) | 97 (43%) | 123 (54%) |
Clinical utility: 8. You are willing to use patient-centered generative AI reports in your practice. | 0 (0%) | 0 (0%) | 2 (1%) | 59 (26%) | 164 (73%) |
