Machine Learning and Wearable Sensors for the Estimation of Biomechanical Variables Outside the Laboratory
| dc.contributor.advisor | Hahn, Michael | |
| dc.contributor.author | Donahue, Seth | |
| dc.date.accessioned | 2022-10-04T20:46:02Z | |
| dc.date.issued | 2022-10-04 | |
| dc.description.abstract | The miniaturization of sensors and their availability for biomechanical analysis outside of the laboratory has opened whole new areas of research. Wearable sensors have been developed to measure ground reaction forces, and inertial measurement units have been developed for the measurement of acceleration and angular velocity. The purpose of this dissertation was to develop methodologies for the measurement and estimation of biomechanical variables, outside of the laboratory. As these sensors can provide vast amounts of data, it is natural to leverage the strengths of machine learning models, which have been used to find patterns in large datasets to assist in the task of estimating biomechanical variables using wearable sensor data as input. This dissertation is divided into five distinct, but related projects all linked to the identification of gait events and machine learning applications for human locomotion data, both in and out of the laboratory. The first two projects were focused on identification of gait events and transitions between locomotion modes, while Projects 3 - 5 were focused on gait event detection and estimation of biomechanical parameters during running outside the laboratory. Project 1: Validation of a supervised machine learning algorithm for steady-state locomotion, and dynamic transitions between those locomotion modes. Project 2: Deployment of an unsupervised machine learning and heuristic gait event detection algorithms for the identification of gait events, across environmentally constrained and internally driven locomotion transitions Projects 3 - 5 resulted in the development of methodologies for biomechanical analysis. We utilized both heuristic and machine learning methodologies for the estimation of biomechanical variables in these scenarios. Project 3: Estimation of gait events and contact times from inertial measures on the foot and the sacrum in a semi-uncontrolled environment. Project 4: Implementation of a recurrent neural network for the estimation of whole ground reaction force waveforms and the calculation of discrete kinetic variables from these waveforms in a semi-uncontrolled environment. Project 5: Synthesis and application of the previous two chapters, gait event detection and estimation of ground reaction force waveforms on data collected in a real-world environment during a 5-mile free run. | en_US |
| dc.description.embargo | 2024-08-09 | |
| dc.identifier.uri | https://hdl.handle.net/1794/27654 | |
| dc.language.iso | en_US | |
| dc.publisher | University of Oregon | |
| dc.rights | All Rights Reserved. | |
| dc.subject | Human Locomotion | en_US |
| dc.subject | Machine Learning | en_US |
| dc.subject | Wearable Sensors | en_US |
| dc.title | Machine Learning and Wearable Sensors for the Estimation of Biomechanical Variables Outside the Laboratory | |
| dc.type | Electronic Thesis or Dissertation | |
| thesis.degree.discipline | Department of Human Physiology | |
| thesis.degree.grantor | University of Oregon | |
| thesis.degree.level | doctoral | |
| thesis.degree.name | Ph.D. |
Files
Original bundle
1 - 1 of 1
Loading...
- Name:
- Donahue_oregon_0171A_13233.pdf
- Size:
- 8.29 MB
- Format:
- Adobe Portable Document Format