Apratim Guha and Atanu Biswas (2008). An overview of modeling techniques for hybrid brain data. Vol. 18, No. 4, 1311-1340.

Back To Index Previous Article Next Article Full Text


Statistica Sinica 18(2008), 1311-1340





AN OVERVIEW OF MODELING TECHNIQUES FOR

HYBRID BRAIN DATA


Apratim Guha$^{1,2}$ and Atanu Biswas$^3$


$^1$National University of Singapore, $^2$University of Birmingham
and $^3$Indian Statistical Institute
Abstract: Constructing models for neuroscience data is a challenging task, more so when the data sets are of hybrid nature, and there exists very little work. The models have to be physiologically meaningful, as well as statistically justifiable. Here we introduce various techniques for fitting a model to bivariate hybrid time series data from the field of neuroscience. As an example, we use a data set on the local field potentials (which is a continuous time series) and nerve cell firings (which is a point process) of anesthetized mice. We extend various available methodologies for modeling nerve cell spike trains to the hybrid set-up, and present a model that has not been previously explored in neuroscience literature. We illustrate the fit of the data set by some Markov chain-based models, some models with crossed dependence including an Inhomogeneous Markov interval type (IMI) model, some ARMA-type models, and also some state space models. We compare the proposals with two existing models. We aim to provide an overview of various possible modeling strategies, and to provide a comparison of the fit and estimation of different models in terms of various standard model selection criteria like AIC and BIC. A detailed simulation study is performed to assess the performance of different models.



Key words and phrases: AIC, BIC, bivariate time series, continuous time series, IMI, jump Markov linear Gaussian system, hybrid process, Markov model, maximum log-likelihood, point process.

Back To Index Previous Article Next Article Full Text