Document Type



Doctor of Philosophy (PhD)


Electrical and Computer Engineering

First Advisor's Name

Malek Adjouadi

First Advisor's Committee Title

Committee chair

Second Advisor's Name

Armando Barreto

Second Advisor's Committee Title

Committee member

Third Advisor's Name

Naphtali Rishe

Third Advisor's Committee Title

Committee member

Fourth Advisor's Name

Mercedes Cabrerizo

Fourth Advisor's Committee Title

Committee member

Fifth Advisor's Name

David A. Loewenstein

Fifth Advisor's Committee Title

Committee member

Sixth Advisor's Name

Jean Andrian

Sixth Advisor's Committee Title

Committee member


engineering, electrical and computer engineering

Date of Defense



Alzheimer's disease (AD) is an irreversible neurodegenerative disorder and a common form of dementia. This research aims to develop machine learning algorithms that diagnose and predict the progression of AD from multimodal heterogonous biomarkers with a focus placed on the early diagnosis. To meet this goal, several machine learning-based methods with their unique characteristics for feature extraction and automated classification, prediction, and visualization have been developed to discern subtle progression trends and predict the trajectory of disease progression.

The methodology envisioned aims to enhance both the multiclass classification accuracy and prediction outcomes by effectively modeling the interplay between the multimodal biomarkers, handle the missing data challenge, and adequately extract all the relevant features that will be fed into the machine learning framework, all in order to understand the subtle changes that happen in the different stages of the disease. This research will also investigate the notion of multitasking to discover how the two processes of multiclass classification and prediction relate to one another in terms of the features they share and whether they could learn from one another for optimizing multiclass classification and prediction accuracy.

This research work also delves into predicting cognitive scores of specific tests over time, using multimodal longitudinal data. The intent is to augment our prospects for analyzing the interplay between the different multimodal features used in the input space to the predicted cognitive scores. Moreover, the power of modality fusion, kernelization, and tensorization have also been investigated to efficiently extract important features hidden in the lower-dimensional feature space without being distracted by those deemed as irrelevant.

With the adage that a picture is worth a thousand words, this dissertation introduces a unique color-coded visualization system with a fully integrated machine learning model for the enhanced diagnosis and prognosis of Alzheimer's disease. The incentive here is to show that through visualization, the challenges imposed by both the variability and interrelatedness of the multimodal features could be overcome. Ultimately, this form of visualization via machine learning informs on the challenges faced with multiclass classification and adds insight into the decision-making process for a diagnosis and prognosis.





Rights Statement

Rights Statement

In Copyright. URI:
This Item is protected by copyright and/or related rights. You are free to use this Item in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you need to obtain permission from the rights-holder(s).