Study design, materials and methods
In 2025, we prospectively compared how 3 expert urologists managed women referring OAB versus the suggestions of AI. OAB had to be the patient's main complaint. All medical evaluations were blind to AI advises. Other physicians reported to ChatGPT Web 3.0 Navigator all the patient’s data evaluated by clinicians and physical examination. Women included could be naïve for OAB, non-responder to first-line treatment, control in responder to therapy, complex cases.
Results
Data of 70 women (mean age 58.7 y.o.) were available: first outpatient visits were 87.1% (61), controls 12.9% (9). Overall, the agreement between physicians and AI was 72.9% (51). At first visits, concordance was found in 70.5% (43/61), disagreement in 18/61 (29.5%). Among controls, 8/9 (88.9%) showed same management. Women with medical indication to pharmacological therapy were 31/70 (44.3%) with agreement in management in 24/31 (77.4%), no concordance in 7/31 (22.6%). In this group, 25/31 (80.6%) were first visits, with disagreement of 24% (6/25) due to different drug class chosen; among the 6/31(19.4%) controls 1 (16.7%) had a different management. Among the 17/70 (24.3%) women with medical indication to OnabotulinToxinA (OBTA), 14/17 (82.4%) were naïve to this treatment and non-responders to previous first-line therapy; of these, 50% (7/14) achieved agreement between medical and IA. In all the 3 women who had a previous OBTA, with good response, there was a complete agreement to continue the same management. Complex cases and women with medical indications to conservative treatment were all first visits. Complex cases were 11/70 (15.7%) with a similar management in 72.7% (8). Conservative management were 15.7% with agreement in 9/11 (81.8%).
Interpretation of results
Among the first visits, conservative management achieved the highest concordance, while OBTA indication as second-line therapy the lowest (50%). This finding can be explained by the possibility of managing these patients in different ways. There was disagreement between physicians and AI In a relevant rate of women undergoing medical indication to pharmacological therapy. This was mainly due to a different type of drug chosen (antimuscarinic vs beta-3), not to a non-pharmacological strategy suggested by AI. A surprisingly higher than expected concordance rate was found in the management of complex cases, which could have numerous variables to consider.