EEG Data Analysis Method And Its Application In Assessing The Emotional Stress State

Stresses for a long time may cause kinds of diseases. The evaluation of the state ofemotional stress is the base of reasonable and effective stress intervention. This paper,aiming at the research on the emotion under stress, made the most of the characteristic thatthe electroencephalogram (EEG) data contained rich emotional information to reach thetarget in combination with complexity, entropy theory and multi-fractal methods. A systemto evaluate the state of emotional stress based on the electroencephalogram had beendesigned and finished.An algorithm was researched in this paper, which focused on the problem ofextracting the EEG signal characters under emotional stress. The degree of randomnesswas described by the KC factor. The complexity and energy distributing could bequantized through the approximate entropy and wavelet entropy in time and frequencydomain respectively. These three characteristic parameters were fused as the emotionalcharacters by the Optimal Support Vector Machine using the genetic algorithm. A total of92groups of EEG signals were collected from14subjects. The results of assessmentshowed that the highest classification accuracy was94.12%, which the average was82.06%. The level of sensitivity to the stress was discrepant for different brain regions.The left hemisphere was more sensitive to stress than the right. The classification accuracyof emotion recognition purposed in this paper could be7.49%higher than the result ofSander Koelstra.The multi-fractal spectrum provided an effective way for nonlinear analysis of EEG.According to the analysis and comparison of the level of signal singularity, the multifractalspectrum of EEG under stress was wider than that without stress. The results suggestedthat the multifractal characteristics of EEG under different stress states were different. Theless the stress be, the weaker the complexity of the brain electrical tends to be. In differenthuman emotional states, the multi-fractal characteristics of EEG were different. Based onthe multi-fractal spectrum feature of emotional EEG, we could effectively classify theemotions. The classification result was28.88%higher than most existing research results. In the Visual Studio2008compiler environment, we developed an emotional stressmeasurement of evaluation system using C#programming mixed with MATALB.According to the relevant EEG data preprocessing, feature extraction and classification,the system would conclude the state of emotional stress.