An RR Intervals Based Arousal Level Index for Emotional IoT

Emotions control our thinking, behavior and actions, and it is of great value for the consumer devices to sensor human emotions and building an emotional Internet of Things. An arousal level index is proposed to extract emotional arousal information from RR intervals, which enable the emotion-aware consumer devices. An SOMFCM clustering method is introduced and an optimized feature set of RR intervals that sensitive to emotions is explored under eight kinds of emotional states, which are no emotion, anger, hate, grief, platonic love, romantic love, joy and reverence. The optimized feature set contains six features, including ＂Mean,＂ ＂Min,＂ ＂2dMean,＂ ＂LF/VLF,＂ ＂NN50＂ and ＂SDNNi.＂ The cluster result shows that RR intervals are sensitive to emotional arousal and the arousal level can be measured by those six features. The change pattern of the six features corresponding to different arousal levels are discussed, and then an emotional arousal level index is proposed based on the cluster result. The proposed arousal level index has a good consistency with the self-assessment of the subject, and is a good indicator of the emotional arousal level. The index makes the emotion-aware capability available to a range of smart devices, and thus to support the emotional Internet of Things.