A probabilistic graphical model that extends Bayesian networks by grouping variables into stages to capture context-specific conditional dependencies.