Estim Ting qua Titat Ve Atural History in Esld

S models are increasingly used to evaluate clinical and public health policies before their implementation; recent examples include models of heart disease, HIV infection, and stroke. As the complexity and clinical realism of these models have increased, the use of standard Markov methods to model cohorts of patients has frequently been replaced with the use of Monte Carlo microsimulation that represents individual patients. Typically, the underlying information used to calibrate these models comes not from actual patient health streams but from statistical models, which produce “average” behaviors for a given patient, based on clinical characteristics. Unfortunately, this averaging tends to smooth the stochastic, sometimes erratic behavior of specific diseases seen in actual patients. Thus, important variation in clinical course may be missed. This is particularly important in diseases such as end-stage liver disease (ESLD), in which patients suffer acute exacerbations and recoveries. Therefore, methods to model these natural histories must incorporate these stochastic changes yet preserve average cohort trends.