Feedback based dynamic environment system for robot-assisted gait training
Date Issued
2023
Author(s)
Rodríguez Muñoz, Arturo Jafet
Advisor(s)
Del Valle Soto, Carolina
González Sánchez, Javier
Type
text::thesis::doctoral thesis
Abstract
Computer-aided rehabilitation therapy has evolved at a staggering rate since the combination
of mechanics, electronics and software was first introduced over two decades ago. Multiple
procedures have been augmented by the use of machines to standardize the therapy and also
to have metrics. Therefore, patients who cannot walk properly due to functional gait disorder
have experienced an evolution from the traditional approaches of using a walking bar to robot-
assisted gait training as shown by Ricklin, Meyer-Heim, and Van Hedel (2018). The main
differences reside in having a more controlled environment where the patient can suffer fewer
accidents, promote muscle memory through proper movements, and measuring the effort done
among other essential data. The Instituto Nacional de Estadística y Geografía (INEGI) creates
multiple census and surveys to map the current situation of Mexico. According to INEGI
(2021) and INEGI (2017) in Mexico in 2020 the total population is approximately 126 million
people where 51.21% are women and 48.78% are men, and 16.53% of the population has
an impairment or limitation. In other words, 20.8 million people have at least one of the
seven most reported impairments which include walking or climbing stairs using their legs.
Unfortunately, robot-assisted gait therapy requires advanced equipment which is expensive.
Consequently, it is not intended to be a personal device but instead a medical device available
only in hospitals and rehabilitation centers. As a result, the amount of people who can benefit
from this technology depends on their location and the availability of each center, so a patient
might have to wait several months before starting therapy. Hence, the importance of having
high efficiency in each session so that each center can help more people.
In this dissertation, we first wondered if having multiple wireless devices on a subject was
possible without suffering from data loss due to interference. The challenge comes from the
possibility of scalability, in other words, introducing a more abundant amount of devices to
collaborate and gather more significant data as time passes. Therefore, we proposed a new
predictive tool and a scheme to address the concerns previously stated. The tool and the
scheme allow the simulation of environments where multiple consumer devices gather informa-
tion from the user regardless of the polling speed or the connection technology employed in a
Wireless Sensor Network (WSN). From the tests we ran, we found that it is possible to have an
environment where many different sensors can feed a system without losing data, which means
that using this simulator, a WSN can be planned for its initial phase or for scaling a previous state. Afterward, we worked on validating that it was possible to use an electroencephalogram
with a virtual reality headset on patients that have a neurological condition. We retrieved
the affective state using a dedicated framework designed for that purpose. During the study,
we found that some of the interpreted values, such as meditation, had a shorter range and
variation for the patients with more neurological affectations. Furthermore, introducing virtual
reality showed an increase of 16% in the efficiency of the session, which was accomplished
with a static scenario. Therefore, we predicted that a dynamic feedback-based scenario could
provide better efficiency than a static scenario. Finally, we worked on the final study, creating
an adaptive feedback system based on adjusting the background music, visual aid brightness,
and overall illumination of a virtual reality environment. The proposed active feedback system
successfully increased the active physical effort done by the users.
of mechanics, electronics and software was first introduced over two decades ago. Multiple
procedures have been augmented by the use of machines to standardize the therapy and also
to have metrics. Therefore, patients who cannot walk properly due to functional gait disorder
have experienced an evolution from the traditional approaches of using a walking bar to robot-
assisted gait training as shown by Ricklin, Meyer-Heim, and Van Hedel (2018). The main
differences reside in having a more controlled environment where the patient can suffer fewer
accidents, promote muscle memory through proper movements, and measuring the effort done
among other essential data. The Instituto Nacional de Estadística y Geografía (INEGI) creates
multiple census and surveys to map the current situation of Mexico. According to INEGI
(2021) and INEGI (2017) in Mexico in 2020 the total population is approximately 126 million
people where 51.21% are women and 48.78% are men, and 16.53% of the population has
an impairment or limitation. In other words, 20.8 million people have at least one of the
seven most reported impairments which include walking or climbing stairs using their legs.
Unfortunately, robot-assisted gait therapy requires advanced equipment which is expensive.
Consequently, it is not intended to be a personal device but instead a medical device available
only in hospitals and rehabilitation centers. As a result, the amount of people who can benefit
from this technology depends on their location and the availability of each center, so a patient
might have to wait several months before starting therapy. Hence, the importance of having
high efficiency in each session so that each center can help more people.
In this dissertation, we first wondered if having multiple wireless devices on a subject was
possible without suffering from data loss due to interference. The challenge comes from the
possibility of scalability, in other words, introducing a more abundant amount of devices to
collaborate and gather more significant data as time passes. Therefore, we proposed a new
predictive tool and a scheme to address the concerns previously stated. The tool and the
scheme allow the simulation of environments where multiple consumer devices gather informa-
tion from the user regardless of the polling speed or the connection technology employed in a
Wireless Sensor Network (WSN). From the tests we ran, we found that it is possible to have an
environment where many different sensors can feed a system without losing data, which means
that using this simulator, a WSN can be planned for its initial phase or for scaling a previous state. Afterward, we worked on validating that it was possible to use an electroencephalogram
with a virtual reality headset on patients that have a neurological condition. We retrieved
the affective state using a dedicated framework designed for that purpose. During the study,
we found that some of the interpreted values, such as meditation, had a shorter range and
variation for the patients with more neurological affectations. Furthermore, introducing virtual
reality showed an increase of 16% in the efficiency of the session, which was accomplished
with a static scenario. Therefore, we predicted that a dynamic feedback-based scenario could
provide better efficiency than a static scenario. Finally, we worked on the final study, creating
an adaptive feedback system based on adjusting the background music, visual aid brightness,
and overall illumination of a virtual reality environment. The proposed active feedback system
successfully increased the active physical effort done by the users.
Subjects
License
Acceso Abierto
URL License
How to cite
Rodríguez Muñoz, A. J. (2023). Feedback based dynamic environment system for robot-assisted gait training. (Tesis de Doctorado). Universidad Panamericana.
Table of contents
Capítulo .1 Methodology, theoretical framework & state of the art -- Capítulo .2 Predictive Tool for Planning the Scaling and Interference of Wireless Sensor Networks -- Capítulo .3 Affective States and Virtual Reality to Improve Gait Rehabilitation: A Preliminary Study -- Capítulo .4 Feedback Based Dynamic Environment System.