作者: Hanieh Bakhshayesh , Sean P. Fitzgibbon , Kenneth J. Pope
DOI: 10.1109/IECBES.2014.7047602
关键词:
摘要: In this paper, we perform the first comparison of a large variety connectivity measures in detecting causal effects among observed interacting systems based on their statistical significance. Well-known estimating direction and strength interdependence between time series are compared: information theoretic measures, model- multivariate frequency domain, phase-based measures. At same phase locking index is used to consider relationship signals, where value implies that response delayed with respect drive at some frequency. The performance tested simulated data from three systems: coupled Henon maps; autoregressive (MVAR) model without EEG as an exogenous input; EEG. No measure was consistently superior. Measures MVAR well when drawn model. Frequency domain have clearly defined band interest. When neither these true, well.