Personalized Rehabilitation Recognition Model upon ANFIS
This study applied the Adaptive Neuro-Fuzzy Inference System (ANFIS) to design a recognition model of personalized rehabilitation. In the model, the user may take a wearable sensor and follow the assigned joint-relax exercise to measure the motions of the upper limbs. The sensor that is embedded wi...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Taiwan Association of Engineering and Technology Innovation
2020-01-01
|
Series: | Proceedings of Engineering and Technology Innovation |
Subjects: | |
Online Access: | http://ojs.imeti.org/index.php/PETI/article/view/3912 |
id |
doaj-f682e35f97c843509601d81fdf5909f6 |
---|---|
record_format |
Article |
spelling |
doaj-f682e35f97c843509601d81fdf5909f62020-11-25T00:57:42ZengTaiwan Association of Engineering and Technology InnovationProceedings of Engineering and Technology Innovation2413-71462518-833X2020-01-0114Personalized Rehabilitation Recognition Model upon ANFISYao-Chiang Kan0Yu-Chieh Kuo1Hsueh-Chun Lin2Department of Communication Engineering, Yuan Ze University, Taoyuan, Taiwan Institute of Health Service Administrations, China Medical University, Taoyuan, TaiwanInstitute of Health Service Administrations, China Medical University, Taoyuan, Taiwan This study applied the Adaptive Neuro-Fuzzy Inference System (ANFIS) to design a recognition model of personalized rehabilitation. In the model, the user may take a wearable sensor and follow the assigned joint-relax exercise to measure the motions of the upper limbs. The sensor that is embedded with the chips of accelerometer, gyroscope, and inclinometer produced the sample datasets due to the exercise schedule of physiotherapy assignment. All motion samples were labeled by arbitrary numbers, which can be identified to the specific motion, for the data training process. A Fuzzy Inference System (FIS) was initially designed by the steps of data pre-processing, featuring, fuzzifying, and ruling Fuzzy logics according to the sample datasets. The FIS was then trained by the ANFIS for optimization by tuning parameters of the features. In testing, the accomplished FIS could recognize the motion features by the defuzzifier that infers the label corresponding to the motion. As a result, the average recognition rate was higher than 90% when the testing motions followed the sampling schedule of the physiotherapy assignment. The model can be applied in the ubiquitous healthcare measurement for health services. The professionals can assess whether the subject obeyed the assigned program or not based on detail motions of the exercise. This approach can be enabled on the trackable interface for the physiatrists to screen the motions of routine rehabilitation. http://ojs.imeti.org/index.php/PETI/article/view/3912ANFISphysiotherapyrehabilitation recognitionsensor-enabled wristbandubiquitous healthcare |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Yao-Chiang Kan Yu-Chieh Kuo Hsueh-Chun Lin |
spellingShingle |
Yao-Chiang Kan Yu-Chieh Kuo Hsueh-Chun Lin Personalized Rehabilitation Recognition Model upon ANFIS Proceedings of Engineering and Technology Innovation ANFIS physiotherapy rehabilitation recognition sensor-enabled wristband ubiquitous healthcare |
author_facet |
Yao-Chiang Kan Yu-Chieh Kuo Hsueh-Chun Lin |
author_sort |
Yao-Chiang Kan |
title |
Personalized Rehabilitation Recognition Model upon ANFIS |
title_short |
Personalized Rehabilitation Recognition Model upon ANFIS |
title_full |
Personalized Rehabilitation Recognition Model upon ANFIS |
title_fullStr |
Personalized Rehabilitation Recognition Model upon ANFIS |
title_full_unstemmed |
Personalized Rehabilitation Recognition Model upon ANFIS |
title_sort |
personalized rehabilitation recognition model upon anfis |
publisher |
Taiwan Association of Engineering and Technology Innovation |
series |
Proceedings of Engineering and Technology Innovation |
issn |
2413-7146 2518-833X |
publishDate |
2020-01-01 |
description |
This study applied the Adaptive Neuro-Fuzzy Inference System (ANFIS) to design a recognition model of personalized rehabilitation. In the model, the user may take a wearable sensor and follow the assigned joint-relax exercise to measure the motions of the upper limbs. The sensor that is embedded with the chips of accelerometer, gyroscope, and inclinometer produced the sample datasets due to the exercise schedule of physiotherapy assignment. All motion samples were labeled by arbitrary numbers, which can be identified to the specific motion, for the data training process. A Fuzzy Inference System (FIS) was initially designed by the steps of data pre-processing, featuring, fuzzifying, and ruling Fuzzy logics according to the sample datasets. The FIS was then trained by the ANFIS for optimization by tuning parameters of the features. In testing, the accomplished FIS could recognize the motion features by the defuzzifier that infers the label corresponding to the motion. As a result, the average recognition rate was higher than 90% when the testing motions followed the sampling schedule of the physiotherapy assignment. The model can be applied in the ubiquitous healthcare measurement for health services. The professionals can assess whether the subject obeyed the assigned program or not based on detail motions of the exercise. This approach can be enabled on the trackable interface for the physiatrists to screen the motions of routine rehabilitation.
|
topic |
ANFIS physiotherapy rehabilitation recognition sensor-enabled wristband ubiquitous healthcare |
url |
http://ojs.imeti.org/index.php/PETI/article/view/3912 |
work_keys_str_mv |
AT yaochiangkan personalizedrehabilitationrecognitionmodeluponanfis AT yuchiehkuo personalizedrehabilitationrecognitionmodeluponanfis AT hsuehchunlin personalizedrehabilitationrecognitionmodeluponanfis |
_version_ |
1725222693809160192 |