Canada Spends Logo

Explorez notre analyse du budget 2026

Grant: Master's Award: Canada Graduate Scholarships

University of Toronto

$17,500.00

Understanding Alzheimer's Disease Heterogeneity using the Subtype and Stage Inference Algorithm

EF 2022-23

Résumé
Alzheimer’s Disease (AD) patient heterogeneity is a key confound to drug design, treatment, and diagnosis. While prior work investigated these different phenotypes (subtypes) using data-driven approaches employing structural MRI, functional MRI or PET data alone, to our knowledge no studies have employed a multimodal approach to map patient heterogeneity, a key objective in the field. In this work, we propose to use the Subtyping, Staging, and Inference (SuStaIn) unsupervised machine learning algorithm to characterize AD heterogeneity, through identification of distinct disease trajectories and rates of progression. We will use a data-driven multimodal approach to train SuStaIn algorithms using amyloid-PET, tau-PET, functional MRI and diffusion tensor imaging from the longitudinal TRIAD dataset. We will train one model using each modality, and one with all modalities combined. We hypothesize that each modality will recover a distinct number of overlapping subtypes as demonstrated by previous literature, and our novel multimodal model will recover its own unique subtypes. We also hypothesize that subtype assignments will stay consistent over time, and each subtype will display distinct cognitive deficits. All the required data, including multiple modalities, timepoints, and cognitive test data have already been collected and preprocessed, and a preliminary model I trained using tau-PET data from the TRIAD dataset was able to recover 4 AD subtypes as demonstrated in the literature. This work will allow us to identify different AD subtypes and their progression patterns to predict changes in pathology and cognition over time, which could lead to personalized treatment strategies, improved diagnostic criteria, and subtype-specific drug development for AD.
Programme
Master's Award: Canada Graduate Scholarships
Bénéficiaire
Emplacement
Ontario, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
CIHR Funding Decisions Database

Tel que publié

Id
486505
Name
Carver Katrina D
Region
Ontario
Country
Canada
Orgtype
Institution Paid; Research Institution
Prcname2
Special Cases - Awards Programs
Projecttitle
Understanding Alzheimer's Disease Heterogeneity using the Subtype and Stage Inference Algorithm
Programtype2
Studentship Programs
Orgnameinp2
University of Toronto
Cihramount2
$17,500
Cihrequipment2
$0
Programname2
Master's Award: Canada Graduate Scholarships
Keyworddelim
Alzheimer'S Disease; Biomarkers; Connectomics; Diffusion Tensor Imaging (Dti); Machine Learning; Magnetic Resonance Imaging (Mri); Multimodal Analyses; Neuroimaging; Positron Emission Tomography (Pet); Unsupervised Learning
Abstract
Alzheimer’s Disease (AD) patient heterogeneity is a key confound to drug design, treatment, and diagnosis. While prior work investigated these different phenotypes (subtypes) using data-driven approaches employing structural MRI, functional MRI or PET data alone, to our knowledge no studies have employed a multimodal approach to map patient heterogeneity, a key objective in the field. In this work, we propose to use the Subtyping, Staging, and Inference (SuStaIn) unsupervised machine learning algorithm to characterize AD heterogeneity, through identification of distinct disease trajectories and rates of progression. We will use a data-driven multimodal approach to train SuStaIn algorithms using amyloid-PET, tau-PET, functional MRI and diffusion tensor imaging from the longitudinal TRIAD dataset. We will train one model using each modality, and one with all modalities combined. We hypothesize that each modality will recover a distinct number of overlapping subtypes as demonstrated by previous literature, and our novel multimodal model will recover its own unique subtypes. We also hypothesize that subtype assignments will stay consistent over time, and each subtype will display distinct cognitive deficits. All the required data, including multiple modalities, timepoints, and cognitive test data have already been collected and preprocessed, and a preliminary model I trained using tau-PET data from the TRIAD dataset was able to recover 4 AD subtypes as demonstrated in the literature. This work will allow us to identify different AD subtypes and their progression patterns to predict changes in pathology and cognition over time, which could lead to personalized treatment strategies, improved diagnostic criteria, and subtype-specific drug development for AD.
Competitiondate
202212
Approvedterm2
1 yr 0 mth
Pinamesdelim
Carver, Katrina D
Approvedterm
1 an 0 mois
Cihrcontribution2
~ $17,500~ $0~