Unboxing iBox: A Critical Appraisal of Its Measurement Properties for Predicting Kidney Allograft Failure

Accurate prediction of kidney allograft failure is key to guiding posttransplant care and stratifying trial participants. The iBox model estimates death-censored graft failure risk at 3, 5, and 7 years posttransplant using demographic, functional, immunologic, and histologic variables. It has been externally validated and to our knowledge, is the first transplant risk score to receive regulatory qualification as a surrogate trial endpoint. This narrative review critically appraises iBox using the Kirschner and Guyatt framework for clinical indices and Steyerberg's framework for predictive models. iBox demonstrates strong sensibility, excellent discrimination (C-index ∼0.81), good calibration, and robust performance across known subgroups. Criterion and construct validity are strong, although several aspects warrant further exploration to support broader implementation. Item reduction and model selection methods were not fully detailed, and key inputs such as biopsy findings and donor-specific antibody levels may vary in availability or be subject to measurement challenges. Some predictors reflect late-stage pathology, which may limit opportunities for early intervention. Formal evaluation of reliability, responsiveness, and interpretability-particularly for longitudinal score changes-remains an important area for future research. By systematically assessing iBox's measurement properties, this review supported its thoughtful implementation and highlighted future research priorities as iBox is integrated into posttransplant care and trial design.