Sensors & Transducers



Vol. 259, Issue 5, October 2022, pp. 29-36





1, * Yasutaka Uchida, 2 Tomoko Funayama, 1 Kazuyoshi Hori, 3 Misako. Yuge,
3 Nobuko Shinozuka and 4 Yoshiaki Kogure



1 Faculty of Life & Environmental Sciences, Teikyo University of Science, Adachi-ku, Tokyo,120-0045, Japan

2 Faculty of Medical Sciences, Teikyo University of Science, Uenohara-shi, Ymanashi, 409-0193, Japan

3 Department of Nursing, Katoriomigawa Medical Center, Katori-shi, Chiba,
289-0332, Japan

4 Professor Emeritus, Teikyo University of Science, Adachi-ku, Tokyo,
120-0045, Japan

Tel.: + 81369101010, fax: + 81369103800

E-mail: uchida@ntu.ac.jp



Received: 3 September 2022 /Accepted: 5 October 2022 /Published: 31 October 2022





Abstract: For improved detection of changes in body conditions, herein, we propose a two-dimensional system, wherein sensors are placed parallel and perpendicular to the direction of walking, based on an already proposed system that employs pressure sensors. The sensors placed parallel to the direction of walking identify the foot that steps on the sensors, and two others pairs of sensors are placed at positions corresponding to the inner and outer sides of the left and right feet, respectively, to accurately detect the foot that steps on the sensors during walking. This improved two-dimensional health monitoring system is applied to a hemodialysis patient known to have a wobble problem before and after treatment. This is performed to obtain effective values as features for machine learning. Using data obtained from the additional sensors, we develop an index that can evaluate a patient's sense of balance. The results indicate that classification is possible based on the walking speed obtained from the pressure sensors installed orthogonally to the direction of travel and the devised balance index. Using these values, the K-means method, which does not use supervised data, can be used to classify the subjects into three regions, and it is demonstrated that changes in gait before and after hemodialysis can be detected, although at an early stage.


Keywords: Health monitoring system, Hemodialysis, Gait balance, Pressure sensors, Machine learning, and K-means method.

_______________________________________________________________________