Physiotherapy Activities Classification with Deep Neural Networks

Düdükçü H. V., Taşkıran M., Dudukcu M. Z.

2022 Innovations in Intelligent Systems and Applications Conference, ASYU 2022, Antalya, Türkiye, 7 - 09 Eylül 2022 identifier

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Doi Numarası: 10.1109/asyu56188.2022.9925547
  • Basıldığı Şehir: Antalya
  • Basıldığı Ülke: Türkiye
  • Anahtar Kelimeler: classification, deep neural networks, inertia sensors, physiotherapy activity, shoulder exercises
  • Yıldız Teknik Üniversitesi Adresli: Evet

Özet

© 2022 IEEE.One of the most important factors affecting the success of the physiotherapy treatment is the participation of the patients in the treatment process. However, in this long process, it is very difficult to monitor the home exercises of the patients. In this study, as the first step of a system that will be developed for the monitoring of home physical therapy exercises, shoulder physiotherapy treatment activities has been classified. In this study, in which 9-axis inertia sensor data was used, the data were processed with three different deep neural networks and the classification performances of the models were compared. Also in this study, by giving time sequences of different lengths as input to the system, the effect of the history window size was also examined, and it was seen that the most successful classification for 11 activity classes was obtained with the 75-time step input sequence using the temporal convolutional networks model.