Japan’s rapidly aging society has led to an increasing number of elderly individuals requiring care, while the available workforce continues to decline, resulting in a critical shortage of caregiving personnel. To address this challenge, we propose a non-wearable bed-monitoring method for posture classification using Radio Frequency Identification (RFID) technology, aiming to prevent fall accidents in care settings. The method leverages the principle that human bodies attenuate radio waves: when a person is positioned between an antenna and tags, the communication status changes depending on their posture. Our system consists of a single antenna mounted above a bed using an aluminum frame and six RFID tags placed beneath the mattress. We classified nine postures―normal sleeping, rolled left/right side, longitudinal sitting, lateral sitting left/right side, terminal sitting left/right side, and left the bed―based solely on the binary communication states without machine learning algorithms. Evaluation experiments were conducted in a laboratory setting with five student participants. The proposed method achieved an average classification accuracy of 96.5%, with individual posture accuracies ranging from 93.2% to 99.5%.
YAMAUCHI et al. (Wed,) studied this question.