Title :
Bayesian nonstationary autoregressive models for biomedical signal analysis
Author :
Cassidy, Michael J. ; Penny, William D.
Author_Institution :
Sobell Dept. of Neurophysiol., Univ. Coll. London, UK
Abstract :
We describe a variational Bayesian algorithm for the estimation of a multivariate autoregressive model with time-varying coefficients that adapt according to a linear dynamical system. The algorithm allows for time and frequency domain characterization of nonstationary multivariate signals and is especially suited to the analysis of event-related data. Results are presented on synthetic data and real electroencephalogram data recorded in event-related desynchronization and photic synchronization scenarios.
Keywords :
Bayes methods; autoregressive processes; electroencephalography; frequency-domain analysis; medical signal processing; physiological models; time series; time-domain analysis; Bayesian nonstationary autoregressive models; EEG analysis; Kalman smoother; biomedical signal analysis; event-related desynchronization; frequency domain characterization; linear dynamical system; photic synchronization scenarios; time domain characterization; time-varying coefficients; variational Bayesian algorithm; Algorithm design and analysis; Bayesian methods; Brain modeling; Electroencephalography; Frequency synchronization; Nervous system; Signal analysis; Signal processing algorithms; Time series analysis; Time varying systems; Algorithms; Bayes Theorem; Computer Simulation; Electroencephalography; Evoked Potentials, Visual; Humans; Likelihood Functions; Linear Models; Models, Statistical; Quality Control; Regression Analysis; Signal Processing, Computer-Assisted; Time Factors;
Journal_Title :
Biomedical Engineering, IEEE Transactions on
DOI :
10.1109/TBME.2002.803511
