Heart Rate Variability Signal Features for Emotion Recognition by Using Principal Component Analysis and Support Vectors Machine

Emotion influences human health significantly. In this pilot study, a movie clips method has been designed to induce 5 kinds of emotion states. 90-sec corresponding ECG signal have been measured in the end of video stimulus. Heart rate variability (HRV) features were extracted from ECG signal by using time-domain, frequency-domain, Poincare, and statistic analysis. Then these HRV features were used to classify different emotion states by support vectors machine (SVM). Also, we used principal component analysis (PCA) to reduce the number of extracted features. Briefly, in the classification for 2 emotion states (positive/negative) and 5 kinds of emotion states, the accuracy of 71.4%, 56.9% are reached, respectively. Compared with other studies of emotion recognition using 2 or more vital signs, the accuracy in this study is lower slightly than other studies (56.9% versus 61.6%). However, using single ECG signal or HRV features is accessible for the daily emotion monitoring. Our results showed the feasibility of daily emotion monitoring by using extracted HRV features and SVM classifier.