turkmath.org

We propose a joint modelling technique for estimating the association between two intensively-measured longitudinal responses: a continuous one and a binary one. Our approach allows us to observe how this relationship varies over time. In addition, the proposed method differs from previous joint modelling techniques by allowing for time-varying effects of the predictors. The motivation for this work comes from a smoking cessation study. We use a joint model of the continuous variable, urge to smoke of a participant, and a binary variable, such as alcohol usage of a participant, to estimate the time-varying association between these responses. Our joint modelling technique can determine both the significance of the relationship and the direction of the relationship. We have shown the asymptotic normality of the estimator and the efficacy of our method is demonstrated via simulation studies.