Research in Statistical Modelling for Longitudinal Data and Survival Data
Research in Statistical Modelling for Longitudinal Data and Survival Data In longitudinal studies, repeated measurements are made on subjects over time and responses within subject are very likely to be correlated. The main aim of longitudinal studies is to investigate how repeated measurements are related to explanatory variables of interest. This sort of data is very common in practice particularly in medicine, public health and economics. Traditional statistical methods for longitudinal data either assume certain parsimonious structures for the within-subject covariance matrix or simply ignore the within-subject correlation for simplicity. This can inevitably lead to misspecification of the within-subject covariance structures, which in turn severely affects statistical inferences. For example, it may result in very inefficient or even biased estimators of model parameters in certain circumstances. Estimation of covariance matrix, however, is rather challenging due to the constraint of its positive definiteness. Professor Pan's research in this area is to
- estimate covariance matrix which may be very large;
- jointly model the mean and covariance structures for longitudinal data;
- carry out variable selection in the joint mean-covariance models;
- model multiple longitudinal processes which are very likely associated.
- non-proportional hazard models;
- competing risk models;
- joint model of longitudinal data and survival data;
- multiple longitudinal processes and competing risks.
Professor Pan is interested in both likelihood-based and Bayesian methods and his research is mainly driven by real problems arising from medical, biological and financial practices. He is also interested in modelling of longitudinal image data and high-dimensional complex data. To date Professor Pan has successfully supervised 9 PhD students and is currently supervising 2 PhD students.
Professor Pan is constantly looking for talented research students.