Estimation of Office Worker’s Emotion Types Using Two-dimensional Model Consisted of Biometric Signals

Abstract

A model that estimates emotion types from bio-signals obtained by wearable monitoring systems was developed. A total of 470 events of strongly conscious emotions (happy, relaxed, sad, and angry) during work hours were recorded in 11 office workers under a total of 911 days of continuous monitoring of pulse rate, pulse rate variability (PRV), skin temperature, body motion, and conversation time with a bracelet-shape wearable sensor. When the four emotion types were developed on the two-dimensional plane consisted of valence and arousal axes, the former axis was composed of the coefficient of variation of high-frequency PRV amplitude and skin temperature, while the latter axis was composed of the frequency variability of PRV respiratory peak and conversation time. Among office workers, two emotional coordinates comprising the Russel's Circumplex model may be estimated by bio-signals obtained by a wearable watch device.