Adaptive functional mixed NHPP models for the analysis of recurrent event panel data

An adaptive semi-parametric model for analyzing longitudinal panel count data is presented. Panel data refers here to data collected as the number of events occurring between specific followup times over a period or disjoint periods of observation of a subject. The counts are assumed to arise from a mixed nonhomogeneous Poisson process where frailties account for heterogeneity common to this type of data. The generating intensity of the counting process is assumed to be a smooth function modeled with penalized splines. A main feature is that the penalization used to control the amount of smoothing, usually assumed to be time homogeneous, is allowed to be time dependent so that the spline can more easily adapt to sharp changes in curvature regimes. Splines are also used to model time dependent covariate effects. Penalized quasi-likelihood (PQL; [Breslow, N.E., Clayton, D.G., 1993. Approximate inference in generalized linear mixed models. Journal of the American Statistical Association 88, 9-25]) is used to derive estimating equations for this adaptive spline model so that only low moment assumptions are required for inference. Both jackknife and bootstrap variance estimators are developed. The finite sample properties of the proposed estimating functions are investigated empirically by simulation. Comparisons with a model assuming a time homogeneous penalty are made. The methods are used in an analysis of data from an experiment to test the effectiveness of pheromones in disrupting the mating pattern of the cherry bark tortrix moth. Recommendations are provided on when the simpler model with a time homogeneous penalty may provide a fair approximation to data and where such an approach will be lacking, calling for the more complicated adaptive methods.