Search

In many experiments, time series data can be collected from multiple units and multiple time series segments can be collected from the same unit. This article introduces a mixed effects Cramér spectral representation which can be used to model the effects of design covariates on the second-order power spectrum while accounting for potential correlations among the time series segments collected from the same unit. The transfer function is composed of a deterministic component to account for the population-average effects and a random component to account for the unit-specific deviations. The resulting log-spectrum has a functional mixed effects representation where both the fixed effects and random effects are functions in the frequency domain. It is shown that, when the replicate-specific spectra are smooth, the log-periodograms converge to a functional mixed effects model. A data-driven iterative estimation procedure is offered for the periodic smoothing spline estimation of the fixed effects, penalized estimation of the functional covariance of the random effects, and unit-specific random effects prediction via the best linear unbiased predictor. [ABSTRACT FROM PUBLISHER]

Hormones serve as regulating signals for many biological processes. In recent years, it was determined that many hormones are secreted in a pulsatile manner and that the pulsatile secretion pattern, in addition to the absolute concentration level, is important in regulating biological processes. Consequently, it is necessary to characterize the latent secretion patterns from measurements of concentration levels. The characterization is complicated by the presence of a biological circadian rhythm. When hormone concentrations are plotted over time, the resultant time series usually exhibits occasional short rises superimposed on a slowly changing baseline. This is a result of a mixture of pulsatile secretions and a circadian rhythm. In this article we present a signal extraction approach to model simultaneously a slowly changing component and a pulsatile component of a time series. A smoothing spline is used to model the baseline, and a multiprocess dynamic linear model is used to model the pulsatile component. An additive structure is assumed, and both components are estimated simultaneously using a multiprocess Kalman filter. The unknown parameters are estimated by approximate maximum likelihood. The locations and amplitudes of the pulses are also estimated as posterior means via the multiprocess Kalman filter. Bayesian confidence intervals can be constructed for the baseline. This approach is found to be robust in simulated data and effective in modeling hormone time series. [ABSTRACT FROM AUTHOR]