Tesi etd-03272023-094740
Link copiato negli appunti
Tipo di tesi
Dottorato
Autore
PERGOLINI, ANDREA
URN
etd-03272023-094740
Titolo
Design and experimental verification of assistive strategies for wearable lower-limb technologies - Applications’ in post-stroke patients and individuals with Parkinson’s disease and limb-amputation
Settore scientifico disciplinare
ING-IND/34
Corso di studi
Istituto di Biorobotica - PHD IN BIOROBOTICA
Commissione
relatore Dott.ssa CREA, SIMONA
Tutor Prof. VITIELLO, NICOLA
Presidente Prof.ssa CASADIO, MAURA
Tutor Prof. VITIELLO, NICOLA
Presidente Prof.ssa CASADIO, MAURA
Parole chiave
- lower-limb exoskeleton
- knee exoskeleton
- locomotion recognition
- rehabilitation
- Parkinson's disease
- stroke
- transfemoral amputees
- biofeedback
- sensorized insoles
Data inizio appello
24/07/2023;
Disponibilità
parziale
Riassunto analitico
Gait is an extraordinarily complex process that requires intact motor and cognitive functions to be performed safely. Neurologic and neurodegenerative diseases commonly compromise motor functions, with gait disorders among the most frequent impairments. Due to the aging of the world population, the occurrence of gait disorders has been increasing significantly; in this context, wearable technologies are emerging as potential allies to improve the quality of life of people with motor disabilities.
This work explored the application of wearable technologies for novel gait training and assistive paradigms, with the objective to provide preliminary evidence of their possible effects with the final end-users. In particular, the work focused on two wearable devices, a wearable sensory biofeedback system and a robotic knee exoskeleton. These devices were tested in case series studies involving people affected by different gait disorders, namely stroke, Parkinson’s disease, and lower-limb amputation.
Concerning the wearable sensory biofeedback device, I conducted two main studies. The first study aimed to evaluate the performance of the device in quantifying kinetic and kinematic parameters of ground-level walking and balance exercises, in people affected by Parkinson’s disease. In particular, the accuracy in detecting gait events and estimating kinetic and kinematics profiles has been assessed through a comparison with gold standard systems used in clinical practice. The results achieved within this study confirmed the suitability of the wearable sensory biofeedback device as a tool for evaluating gait’s temporal parameters in ecological conditions (with absolute errors in stance phase estimation lower than 2% of the gait cycle) and for assessing ground-reaction force profiles. The second study was a case series aiming to investigating whether the wearable sensory biofeedback apparatus could elicit improvements in gait determinants in people affected by Parkinson’s disease. After three days of training with the device, results reported higher walking distance, on average +10% increment during a 2-minutes walking test, when using the device compared to pre-training. Additionally, results indicated trends toward higher stride length, cadence, gait speed, and double-support phase duration when using the device, compared to walking without it. Future experiments are still needed to investigate long terms effects and to evaluate the required training period to obtain clinically significant results.
Concerning the unilateral knee exoskeleton, I conducted an experimental trial to investigate the control performance with the subjects’ hemiparetic gait in steady-state walking on a treadmill. Secondly, the experimental trial allowed to assess the effects of assisting the knee flexion-extension of post-stroke subjects with mild-to-moderate impairments. Results on the exoskeleton control have proved its capabilities to estimate the gait phase in real-time to assist the subjects with an error lower than 2.5 % of the gait cycle (against ± 4 % previously suggested by the literature as a benchmark error). Furthermore, the results suggested that the flexion-extension assistance could improve the knee kinematics of the post-stroke subjects in steady-state walking. The effects of the exoskeleton assistance resulted in a clinically significant increment of the range of motion of the paretic knee in the sagittal plane (> 16 deg) in all participants.
Aiming to improve further the knee exoskeleton control with different locomotion tasks and transitory states, this thesis work included an initial attempt at a novel middle-level controller. The novel controller was initially tested offline, with promising results, on a 6 healthy subjects dataset performing a continuous circuit of multiple locomotion tasks.
Lastly, this thesis presents the design and development of a novel locomotion recognition algorithm that exploits the biomechanical peculiarities of the two lower-limb sides using two pairs of independent classifiers (one pair per side). The algorithm was designed to be easily adapted to unilateral exoskeletons and could be potentially implemented on the knee exoskeleton. Even though, in this thesis work, it was preliminarily tested using a powered bilateral hip exoskeleton. Subject-dependent algorithm models were tested in real-time with lower-limb transfemoral amputees achieving a median recognition accuracy of 94.8% (across subjects).
Overall, this work contributed to advancing the state of the art on wearable sensory biofeedback devices and lower-limb exoskeletons with preliminary studies with the final end-users. The performed studies provided insights on the feasibility and potential advantage of using wearable sensory biofeedback devices in the gait training paradigm with people affected by Parkinson's disease, as well as exploiting powered wearable exoskeletons for enhancing gait in post-stroke subjects and transfemoral amputees.
This work explored the application of wearable technologies for novel gait training and assistive paradigms, with the objective to provide preliminary evidence of their possible effects with the final end-users. In particular, the work focused on two wearable devices, a wearable sensory biofeedback system and a robotic knee exoskeleton. These devices were tested in case series studies involving people affected by different gait disorders, namely stroke, Parkinson’s disease, and lower-limb amputation.
Concerning the wearable sensory biofeedback device, I conducted two main studies. The first study aimed to evaluate the performance of the device in quantifying kinetic and kinematic parameters of ground-level walking and balance exercises, in people affected by Parkinson’s disease. In particular, the accuracy in detecting gait events and estimating kinetic and kinematics profiles has been assessed through a comparison with gold standard systems used in clinical practice. The results achieved within this study confirmed the suitability of the wearable sensory biofeedback device as a tool for evaluating gait’s temporal parameters in ecological conditions (with absolute errors in stance phase estimation lower than 2% of the gait cycle) and for assessing ground-reaction force profiles. The second study was a case series aiming to investigating whether the wearable sensory biofeedback apparatus could elicit improvements in gait determinants in people affected by Parkinson’s disease. After three days of training with the device, results reported higher walking distance, on average +10% increment during a 2-minutes walking test, when using the device compared to pre-training. Additionally, results indicated trends toward higher stride length, cadence, gait speed, and double-support phase duration when using the device, compared to walking without it. Future experiments are still needed to investigate long terms effects and to evaluate the required training period to obtain clinically significant results.
Concerning the unilateral knee exoskeleton, I conducted an experimental trial to investigate the control performance with the subjects’ hemiparetic gait in steady-state walking on a treadmill. Secondly, the experimental trial allowed to assess the effects of assisting the knee flexion-extension of post-stroke subjects with mild-to-moderate impairments. Results on the exoskeleton control have proved its capabilities to estimate the gait phase in real-time to assist the subjects with an error lower than 2.5 % of the gait cycle (against ± 4 % previously suggested by the literature as a benchmark error). Furthermore, the results suggested that the flexion-extension assistance could improve the knee kinematics of the post-stroke subjects in steady-state walking. The effects of the exoskeleton assistance resulted in a clinically significant increment of the range of motion of the paretic knee in the sagittal plane (> 16 deg) in all participants.
Aiming to improve further the knee exoskeleton control with different locomotion tasks and transitory states, this thesis work included an initial attempt at a novel middle-level controller. The novel controller was initially tested offline, with promising results, on a 6 healthy subjects dataset performing a continuous circuit of multiple locomotion tasks.
Lastly, this thesis presents the design and development of a novel locomotion recognition algorithm that exploits the biomechanical peculiarities of the two lower-limb sides using two pairs of independent classifiers (one pair per side). The algorithm was designed to be easily adapted to unilateral exoskeletons and could be potentially implemented on the knee exoskeleton. Even though, in this thesis work, it was preliminarily tested using a powered bilateral hip exoskeleton. Subject-dependent algorithm models were tested in real-time with lower-limb transfemoral amputees achieving a median recognition accuracy of 94.8% (across subjects).
Overall, this work contributed to advancing the state of the art on wearable sensory biofeedback devices and lower-limb exoskeletons with preliminary studies with the final end-users. The performed studies provided insights on the feasibility and potential advantage of using wearable sensory biofeedback devices in the gait training paradigm with people affected by Parkinson's disease, as well as exploiting powered wearable exoskeletons for enhancing gait in post-stroke subjects and transfemoral amputees.
File
| Nome file | Dimensione |
|---|---|
Ci sono 1 file riservati su richiesta dell'autore. |
|