A semi-parametric Bayesian dynamic hurdle model with an application to the health and retirement study

All developed countries are facing the problem of providing affordable and high quality healthcare in recent years. This is due to combination an ageing population technological advancement health science which leads increased life expectancy. will affect future use hospital inpatient outpatient services turn place a significant stress on economy since most medical for elderly apportioned funded under national system. Thus, understanding demand other key factors influencing crucial better serve citizens. Hospital admission considered be proxy healthcare, especially context populations as experienced globally. However, modeling admissions, although very important, often complicated by zero-inflation, covariates with time-varying effects, necessity borrowing information across individuals. Additionally, rate admissions might differ between group individuals who have been hospitalized before yet hospitalized. Also when clustered based their baseline self-assessed status, distribution its relation predictors may quite different within groups. In this paper we propose unified Bayesian dynamic hurdle model accommodates these features data semi-parametric approach. We analyze collected United States Health Retirement Study varies Simulation studies performed assessing usefulness proposed model.