Table 2. Characteristics of prediction models developed to estimate the risk of disease reclassification during AS.

Prediction model	Statistical technique	Development population	Endpoint
Canary-PASS	Generalized estimating equations (GEE)	859 patients from the Canary-PASS cohort	GS ≥7 or ≥34% of tumor-positive cores
Johns Hopkins	Bayesian hierarchical latent class model	964 patients from Johns Hopkins	GS ≥7 at radical prostatectomy
PRIAS	Joint model*	5,624 patients from the PRIAS project	Time until GS ≥7 at biopsy

*, a combination of a mixed effect model and a survival model. AS, active surveillance.