Biostatistics Seminar Series: Jing Huang, PhD

Postdoctoral Researcher Division of Biostatistics Department of Biostatistics, Epidemiology and Informatics University of Pennsylvania Perelman School of Medicine Abstract:  Nowadays, biomedical research are facilitated with massive healthcare data collected from different sources with unique features, e.g., data from safety surveillance monitoring systems and data from electronic health records. Traditional likelihood methods sometimes fail under such settings, due to the large scale of data and the complex data generating mechanism. Compared to traditional likelihood methods, composite likelihood methods may be less statistically efficient, but they can be formulated with more computational convenience and robustness to model misspecification. In this talk, I will present a serial of work to show how composite likelihood functions can be used to generate evidence from large-scale healthcare data. I will demonstrate a composite likelihood approach for monitoring vaccine safety using data from a national spontaneous surveillance system. I will show the statistical theory of such a formulation, and introduce a new framework of specification tests to evaluate the goodness-of-fit for composite likelihood inference.