便座心電図を用いた個人識別の基礎的研究

抄録

In order to achieve health management and disease prevention, many studies of toilets with sensors have been conducted to measure daily physiological information. These studies require person identification, assuming that several people use the same toilet. In this study, we focused on the identification of a person while sitting on a toilet seat. Three electrodes were attached to the toilet seat, and identification was performed using an electrocardiogram recorded from both thighs. The aim of this study was to investigate the usefulness of the proposed person identification method. We calculated the identification rate and equal error rate (EER) for the proposed person identification method. Fifty thigh electrocardiogram samples were obtained from 5 subjects. In addition, an electrocardiogram was simultaneously recorded from the chest of each subject. The R wave, S wave, and T wave were clearly obtained from both electrocardiograms. The peak/negative peak times of R, S, and T waves recorded from both thighs correlated with those measured simultaneously from the chest. Eleven features including amplitude and interval were extracted from the thigh electrocardiogram. Discriminant analysis using Mahalanobis distance was performed to calculate the identification rate. The average identification rate was found to be 82%. Furthermore, the EER, which is an accuracy evaluation index for person identification, was calculated. The average of EER was found to be 15.8%.