Validate prospective response prediction outside the evaluated cohorts

Validate whether scDEFT prospectively predicts treatment response in patients outside the three inflammatory bowel disease cohorts used for model development and evaluation.

Background

The three-cohort analysis reports a prospective AUROC of 0.70 for responder prediction from pre-treatment cells. Because the result is based on a limited set of cohorts, the authors explicitly identify external confirmation in additional patients as unresolved future work. This is an external-validity question rather than a claim that the existing result is already established broadly.