In recent years, the number of patients with lifestyle-related diseases has been increasing. Some lifestyle-related diseases progress without symptoms and regular checkups, thus, continuous recording of individual health information are considered important for prevention. Therefore, systems that automatically record health information such as blood pressure, body temperature, composition of urine and feces are being developed by installing sensors in the toilet used in daily life. However, it is necessary to identify each individual toilet user because toilets are expected to be used by multiple people. Therefore, Kurahashi et al(1). identified individuals using an angular velocity sensor installed inside a toilet paper core. However, it uses only feature values such as the amount of toilet paper used and the maximum rewind speed, and identification results using the K-NearestNeighbour method (K=1) were only reported. Therefore, we analyze toilet paper winding motions to examine features effective for personal identification, and evaluate the identification accuracy using other classification methods. As a result, we confirmed that the identification accuracy was improved by adding new features on the toilet paper winding motion.
Masaki et al. (2026) studied this question.