Brain-computer interface (BCI) systems identify brain signals through extracting features from them. In view of the limitations of the autoregressive model feature extraction method and the traditional principal component analysis to deal with the multichannel signals, this paper presents a multichannel feature extraction method that multivariate autoregressive (MVAR) model combined with the multiple-linear principal component analysis (MPCA), and used for magnetoencephalography (MEG) signals and electroencephalograph (EEG) signals recognition. Firstly, we calculated the MVAR model coefficient matrix of the MEG/EEG signals using this method, and then reduced the dimensions to a lower one, using MPCA. Finally, we recognized brain signals by Bayes Classifier. The key innovation we introduced in our investigation showed that we extended the traditional single-channel feature extraction method to the case of multi-channel one. We then carried out the experiments using the data groups ofⅣ_ⅢandⅣ_Ⅰ. The experimental results proved that the method proposed in this paper was feasible.
Heart rate variability (HRV) analysis technology based on an autoregressive (AR) model is widely used in the assessment of autonomic nervous system function. The order of AR models has important influence on the accuracy of HRV analysis. This article presents a method to determine the optimum order of AR models. After acquiring the ECG signal of 46 healthy adults in their natural breathing state and extracting the beat-to-beat intervals (RRI) in the ECG, we used two criteria, i.e. final prediction error (FPE ) criterion to estimate the optimum model order for AR models, and prediction error whiteness test to decide the reliability of the model. We compared the frequency domain parameters including total power, power in high frequency (HF), power in low frequency (LF), LF power in normalized units and ratio of LF/HF of our HRV analysis to the results of Kubios-HRV. The results showed that the correlation coefficients of the five parameters between our methods and Kubios-HRV were greater than 0.95, and the Bland-Altman plot of the parameters was in the consistent band. The results indicate that the optimization algorithm of HRV analysis based on AR models proposed in this paper can obtain accurate results, and the results of this algorithm has good coherence with those of the Kubios-HRV software in HRV analysis.