Mehdipour Ghazi, Mostafa;
(2021)
Robust Modeling and Prediction of Disease Progression Using Machine Learning.
Doctoral thesis (Ph.D), UCL (University College London).
Preview |
Text
Thesis-Mostafa.pdf - Accepted Version Download (6MB) | Preview |
Abstract
This work studies modeling the progression of Alzheimer’s disease using a parametric method robust to outliers and missing data and a nonparametric method robust to missing values and training instabilities. The proposed parametric method linearly maps the individual’s age to a disease progression score (DPS) and jointly fits constrained generalized logistic functions to the longitudinal dynamics of biomarkers as functions of the DPS using M-estimation. The proposed nonparametric method applies a generalized training rule based on normalizing the input and loss to the number of available data points to the long short-term memory (LSTM) recurrent neural networks to handle missing input and target values. Moreover, a robust initialization method is developed to address the training instability in LSTM networks based on a scaled random initialization of the network weights, aiming at preserving the variance of the network input and output in the same range. Both proposed methods are evaluated on data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) for robust modeling of volumetric magnetic resonance imaging (MRI) and positron emission tomography (PET) biomarkers, cerebrospinal fluid (CSF) measurements, as well as cognitive tests, and are compared to the state-of-the-art methods. The obtained results show that the proposed parametric model outperforms almost all state-of-the-art parametric methods in predicting biomarker values and classifying clinical status, and it generalizes well when applied to independent data from the National Alzheimer’s Coordinating Center (NACC). Additionally, the proposed generalized training rule for deep neural networks achieves superior results to standard LSTMs using data imputation before training, especially when applied to data with lower rates of missing values. A comprehensive analysis of the proposed methods in neurodegenerative disease progression modeling reveals that the proposed nonparametric method performs better than the proposed parametric method in predicting biomarker values, while the parametric method works significantly better in clinical status classification.
| Type: | Thesis (Doctoral) |
|---|---|
| Qualification: | Ph.D |
| Title: | Robust Modeling and Prediction of Disease Progression Using Machine Learning |
| Event: | UCL (University College London) |
| Open access status: | An open access version is available from UCL Discovery |
| Language: | English |
| Additional information: | Copyright © The Author 2021. Original content in this thesis is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) Licence (https://creativecommons.org/licenses/by-nc/4.0/). Any third-party copyright material present remains the property of its respective owner(s) and is licensed under its existing terms. Access may initially be restricted at the author’s request. |
| UCL classification: | UCL UCL > Provost and Vice Provost Offices UCL > Provost and Vice Provost Offices > UCL BEAMS UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Engineering Science UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Engineering Science > Dept of Med Phys and Biomedical Eng |
| URI: | https://discovery.ucl.ac.uk/id/eprint/10130072 |
Archive Staff Only
 |
View Item |
