Bisele, M ORCID: https://orcid.org/0000-0002-3785-0020, 2018. Assessment and understanding of unilateral trans-tibial amputee gait using principal component analysis and discriminant function analysis. PhD, Nottingham Trent University.
Preview |
Text
Maria Bisele - Final PhD Thesis.pdf - Published version Download (21MB) | Preview |
Abstract
The general aim of this thesis was to develop analytical techniques for the assessment and understanding of lower-limb amputee (LLA) gait. The number of individuals with lower limb amputation (LLA) worldwide is growing and being able to optimise rehabilitation and prosthetic prescriptions are becoming more important. Gait analysis may be able to inform these processes, in particular at the individual level.
In study one, a machine learning algorithm was developed and optimised using Principal Component Analysis (PCA) and Discriminant Function Analysis (DFA) to distinguish between barefoot and shod running. An iterative process was used to optimise the algorithm, exploring all possible iterations of ten individuals out of twenty, finding the combination of people with the greatest generic features and thus the lowest error rate for classification. The outcome showed 93.5% classification accuracy between barefoot and shod running. This study demonstrated that an iteration procedure could optimise a machine learning algorithm to overcome the issues of overfitting, which is particularly useful when working with a small sample size as is common in gait analysis.
In study two, PCA and DFA were used to identify differences between the gait of individuals with unilateral trans-tibial amputation (UTTA) and able-bodied individuals. Different approaches were explored, establishing that PCA conducted on normalised temporal-waveforms yielded the best outcome. Results revealed that UTTA and able-bodied gait differed with regards to certain biomechanical variables, providing a better understanding of LLA function. Although differences between individuals with LLA and able-bodied individuals have previously been investigated, this study demonstrates that using multivariate statistical analyses a vast number of variables can be investigated simultaneously, identifying the hierarchy of variables and thus which need to be targeted during treatment.
Clinical diagnosis is based on individual patients, thus in study three PCA was used to determine whether one individual with a UTTA displayed unique gait characteristics when compared to a group of able-bodied individuals. Both covariance and correlation matrices were used during PCA, providing information about variation and magnitude of the data, respectively. Results revealed that each individual with UTTA has subject-specific gait characteristics, which highlights that this method can be used to identify variables which can be targeted during treatment.
In the fourth and final study, PCA was used to understand the effects of attempted symmetry on dynamic stability of individuals with UTTA. Although in rehabilitation symmetrical gait is often sought for since asymmetrical gait is said to cause long term adverse effects, results revealed that asymmetry might be playing a functional role and in fact aids better stability in UTTA gait. This outcome may suggest that after a certain symmetry has been reached, the target of rehabilitation may need to be reconsidered to aim for better stability.
In conclusion, multivariate statistical analysis could be used to assess and understand LLA function. In a clinical setting, the ability to identify important variables during a task, particularly at patient-specific level has the potential to improve the development of treatment recommendations. Prosthetic prescription and rehabilitation processes can be tailored and in turn the outcome may be more successful which could increase the likelihood of independent living of patients and therefore improve the quality of life of individuals with LLA.
Item Type: | Thesis |
---|---|
Creators: | Bisele, M. |
Date: | June 2018 |
Rights: | This work is the intellectual property of the author. You may copy up to 5% of this work for private study, or personal, non-commercial research. Any re-use of the information contained within this document should be fully referenced, quoting the author, title, university, degree level and pagination. Queries or requests for any other use, or if a more substantial copy is required, should be directed to the owner(s) of the Intellectual Property Rights. |
Divisions: | Schools > School of Science and Technology |
Record created by: | Linda Sullivan |
Date Added: | 01 May 2019 12:34 |
Last Modified: | 20 Aug 2020 15:25 |
URI: | https://irep.ntu.ac.uk/id/eprint/36372 |
Actions (login required)
Edit View |
Statistics
Views
Views per month over past year
Downloads
Downloads per month over past year