This study tested an artificial intelligence (AI) system for scoring ulcerative colitis during endoscopy. It compared the AI system to the standard Mayo endoscopic score in 387 patients from a clinical trial. The main goal was to see how well each scoring method matched patient symptoms and quality of life.
The AI system showed better agreement with patient-reported symptoms and quality of life measures than the standard score. Patients with lower AI scores were more likely to be in symptomatic remission. The study also found that a low AI score cutoff had good agreement with remission status.
No safety concerns were reported for this analysis, as it focused on scoring agreement. The main reason to be careful is that this is an observational analysis within a larger trial; it shows an association, not that the AI system causes better outcomes. The results are from a phase 3 trial, so the certainty is high for the reported agreement measures.
Readers should understand that this study suggests AI scoring may align better with how patients feel, but it does not show that using AI changes treatment decisions or improves health outcomes. More research is needed to see if this scoring method leads to better patient care.