An AI-powered Red Density (RD) Score that measures colonic mucosal redness at the pixel level predicted long-term clinical outcomes in patients with ulcerative colitis who were in remission. In the prospective PROCEED-UC study of 176 patients across 10 academic centers, those who later experienced clinical failure had significantly higher baseline RD scores than those who remained in remission (52.94 vs. 32.71; P=0.033).
A baseline RD score of 63 or higher was associated with clinical failure, with the tool achieving an AUC of 0.82 (P<0.0001). The score also showed moderate correlations with established histologic measures of inflammation, suggesting that AI-based image analysis could potentially identify subclinical inflammation and flag patients at risk of relapse—even while they appear clinically well.