borderline significanceP = 0.055
In the development cohort, the Hosmer–Lemeshow test yielded a non-significant P -value of 0.299, while the temporal validation cohort showed borderline significance ( P = 0.055), indicating acceptable model calibration.