Related Experiment Video
Updated: Jul 15, 2026
Author Spotlight: Automated Deep Brain Stimulation for Parkinson's Disease - Exploring the Possibilities and Challenges of Home Monitoring
Published on: July 14, 2023
Identification of People in a Household Using Ballistocardiography Signals Through Deep Learning
Karin Takahashi1, Yoshinobu Tanno1, Hitoshi Ueno1
1Faculty of Information Design, Tokyo Information Design Professional University, Edogawa-ku, Tokyo 132-0034, Japan.
Background:
Various sensor technologies have been developed to monitor the health of older adults; however, most of them require attachment to the skin. This study aimed to develop a health monitoring system, using a non-adhesive, non-invasive polyvinylidene difluoride piezoelectric sensor, with the patient being able to lead a normal daily life without being conscious of the sensor. The vibration signal from the human body surface obtained by the piezoelectric sensor, which is a ballistocardiography signal, contains information on the person's heart and respiratory rates. We propose a method that enables individual identification based on the characteristics of the frequency components of the signal.
Methods:
Signals from ten subjects were acquired and a neural network was constructed, trained, and tested using 252 cases to identify five individuals, based on assuming the number of people in a household.
Results:
The classification probability and accuracy rate were obtained for all 252 cases, and good classification rates were obtained in almost all cases.
Conclusions:
Although it will be necessary to consider daily changes in such signals in the future, the system had good identification accuracy when five individuals were identified.
