A noninvasive sensor system for discriminating mobility pattern on a bed

Abstract

In this paper, we propose a noninvasive sensor system for discriminating mobility pattern of a resident on the bed without inconvenience. The proposed system consists of a thin and wide film style of piezoelectric force senor, a signal processing board, and data collecting program. There are four different types of motion that were simulated by non-patient volunteers. About 10,000 experimental motions of subjects were performed. Sensor data by human motions were collected and preprocessed by a moving average filter, transformed by FFT, and classified by the k-NN algorithm with k = 1. The experiment yielded the overall discrimination rate of 89.4%. The proposed system will contribute to differentiating mobility pattern on a bed and distinguishing the physical characteristics of a person.