Wireless Mobile Communication and Healthcare. 8th EAI International Conference, MobiHealth 2019, Dublin, Ireland, November 14-15, 2019, Proceedings

Research Article

Preliminary Investigation on Band Tightness Estimation of Wrist-Worn Devices Using Inertial Sensors

  • @INPROCEEDINGS{10.1007/978-3-030-49289-2_20,
        author={Masayuki Hayashi and Hiroki Yoshikawa and Akira Uchiyama and Teruo Higashino},
        title={Preliminary Investigation on Band Tightness Estimation of Wrist-Worn Devices Using Inertial Sensors},
        proceedings={Wireless Mobile Communication and Healthcare. 8th  EAI International Conference, MobiHealth 2019, Dublin, Ireland, November 14-15, 2019, Proceedings},
        proceedings_a={MOBIHEALTH},
        year={2020},
        month={6},
        keywords={Wrist-worn device Inertial sensor Machine learning Tightness estimation},
        doi={10.1007/978-3-030-49289-2_20}
    }
    
  • Masayuki Hayashi
    Hiroki Yoshikawa
    Akira Uchiyama
    Teruo Higashino
    Year: 2020
    Preliminary Investigation on Band Tightness Estimation of Wrist-Worn Devices Using Inertial Sensors
    MOBIHEALTH
    Springer
    DOI: 10.1007/978-3-030-49289-2_20
Masayuki Hayashi1,*, Hiroki Yoshikawa1,*, Akira Uchiyama1,*, Teruo Higashino1,*
  • 1: Osaka University
*Contact email: m-hayashi@ist.osaka-u.ac.jp, h-yoshikawa@ist.osaka-u.ac.jp, uchiyama@ist.osaka-u.ac.jp, higashino@ist.osaka-u.ac.jp

Abstract

Nowadays, wearable devices enable us to collect biological data from a massive number of people. However, the reliability of the collected data varies due to various factors such as band tightness and incorrect attachment. In this paper, we investigate the band tightness estimation by using an inertial sensor of a wrist-worn device. First, we analyze the relationship between the band tightness and the data reliability through a preliminary experiment. Then, we design the band tightness estimation as a classification problem based on frequency domain features. The evaluation results show the effectiveness of the frequency domain features, achieving the accuracy of 81.7% for the 3-class band tightness classification.