Functional survival forests for multivariate longitudinal outcomes: Dynamic prediction of Alzheimer’s disease progression

The random survival forest (RSF) is a non-parametric alternative to the Cox proportional hazards model in modeling time-to-event data. In this article, we developed a modeling framework to incorporate multivariate longitudinal data in the model building process to enhance the predictive performance of RSF. To extract the essential features of the multivariate longitudinal outcomes, two methods were adopted and compared: multivariate functional principal component analysis and multivariate fast covariance estimation for sparse functional data. These resulting features, which capture the trajectories of the multiple longitudinal outcomes, are then included as time-independent predictors in the subsequent RSF model. This non-parametric modeling framework, denoted as functional survival forests, is better at capturing the various trends in both the longitudinal outcomes and the survival model which may be difficult to model using only parametric approaches. These advantages are demonstrated through simulations and applications to the Alzheimer’s Disease Neuroimaging Initiative.

Download Full-text

Characterizing the spatiotemporal variability of Alzheimer's disease pathology

10.1101/2020.08.20.20176883 ◽

2020 ◽

Author(s):

Jacob W Vogel ◽

Alexandra L Young ◽

Neil P Oxtoby ◽

Ruben Smith ◽

Rik Ossenkoppele ◽

...

Keyword(s):

Alzheimer’S Disease ◽

Alzheimer's Disease ◽

Medial Temporal Lobe ◽

Spatiotemporal Variability ◽

Cognitive Profiles ◽

Tau Pathology ◽

Modeling Framework ◽

Longitudinal Outcomes ◽

Younger Age ◽

Clinical Variants

Alzheimer's disease (AD) is characterized by the progressive spread of tau pathology throughout the cerebral cortex. The pattern of spread is thought to be fairly consistent across individuals, though more recent work has demonstrated substantial variability in the AD population that is often associated with distinct clinical phenotypes. Still, a systematic, unbiased, whole-brain characterization of spatiotemporal variation in tau deposition in AD is lacking. We analyzed 1612 tau-PET scans and applied to this sample a disease progression modeling framework designed to identify spatiotemporal trajectories of pathological progression. We identified four distinct trajectories of tau progression, ranging in prevalence from 18-33\%, with no one progession predominating. We replicated previously described limbic-predominant and medial temporal lobe-sparing variants, while also discovering posterior and lateral temporal subtypes resembling atypical clinical variants of AD. These "subtypes" were stable during longitudinal follow-up, and could be replicated in a separate sample using a different radiotracer. The subtypes presented with distinct demographic and cognitive profiles and differing longitudinal outcomes, however, no "typical" variant predominated. Across all subtypes, younger age was related to worse cognition and more rapid tau accumulation. Additionally, network diffusion models implicated that pathology originates and spreads through distinct corticolimbic in the different subtypes. Together, our results suggest variation in tau pathology is common and systematic, perhaps warranting a re-examination of the notion of "typical AD", and a revisiting of tau pathological staging.

Download Full-text

VariFunNet, an integrated multiscale modeling framework to study the effects of rare non-coding variants in genome-wide association studies: Applied to Alzheimer's disease

2017 IEEE International Conference on Bioinformatics and Biomedicine (BIBM) ◽

10.1109/bibm.2017.8217995 ◽

2017 ◽

Cited By ~ 3

Author(s):

Qiao Liu ◽

Chen Chen ◽

Annie Gao ◽

Hang Hang Tong ◽

Lei Xie

Keyword(s):

Alzheimer’S Disease ◽

Alzheimer's Disease ◽

Multiscale Modeling ◽

Association Studies ◽

Genome Wide Association ◽

Genome Wide Association Studies ◽

Modeling Framework ◽

Genome Wide ◽

Multiscale Modeling Framework ◽

Coding Variants

Download Full-text

Early diagnosis of Alzheimer’s disease on ADNI data using novel longitudinal score based on functional principal component analysis

Journal of Medical Imaging ◽

10.1117/1.jmi.8.2.024502 ◽

2021 ◽

Vol 8 (02) ◽

Author(s):

Haolun Shi ◽

Da Ma ◽

Yunlong Nie ◽

Mirza Faisal Beg ◽

Jian Pei ◽

...

Keyword(s):

Alzheimer’S Disease ◽

Alzheimer's Disease ◽

Principal Component Analysis ◽

Early Diagnosis ◽

Principal Component ◽

Component Analysis ◽

Functional Principal Component Analysis ◽

Functional Principal Component

Download Full-text

Changes in Metabolic Syndrome Status and Risk of Dementia

Journal of Clinical Medicine ◽

10.3390/jcm9010122 ◽

2020 ◽

Vol 9 (1) ◽

pp. 122 ◽

Cited By ~ 2

Author(s):

Ji Eun Lee ◽

Dong Wook Shin ◽

Kyungdo Han ◽

Dahye Kim ◽

Jung Eun Yoo ◽

...

Keyword(s):

Alzheimer’S Disease ◽

Blood Pressure ◽

Alzheimer's Disease ◽

Metabolic Syndrome ◽

Vascular Dementia ◽

Proportional Hazards ◽

Proportional Hazards Model ◽

Screening Program ◽

Cox Proportional Hazards ◽

Cox Proportional Hazards Model

This study investigated the effects of changes in metabolic syndrome (MS) status and each component on subsequent dementia occurrence. The study population was participants of a biennial National Health Screening Program in 2009–2010 and 2011–2012 in Korea. Participants were divided into four groups according to change in MS status during the two-year interval screening: sustained normal, worsened (normal to MS), improved (MS to normal), and sustained MS group. Risk of dementia among the groups was estimated from the second screening date to 31 December 2016 using a Cox proportional hazards model. A total of 4,106,590 participants were included. The mean follow-up was 4.9 years. Compared to the sustained normal group, adjusted hazard ratios (aHR) (95% confidence interval) were 1.11 (1.08–1.13) for total dementia, 1.08 (1.05–1.11) for Alzheimer’s disease, and 1.20 (1.13–1.28) for vascular dementia in the worsened group; 1.12 (1.10–1.15), 1.10 (1.07–1.13), and 1.19 (1.12–1.27) for the improved group; and 1.18 (1.16–1.20), 1.13 (1.11–1.15), and 1.38 (1.32–1.44) for the sustained MS group. Normalization of MS lowered the risk of all dementia types; total dementia (aHR 1.18 versus 1.12), Alzheimer’s disease (1.13 versus 1.10), and vascular dementia (1.38 versus 1.19). Among MS components, fasting glucose and blood pressure showed more impact. In conclusion, changes in MS status were associated with the risk of dementia. Strategies to improve MS, especially hyperglycemia and blood pressure, may help to prevent dementia.

Download Full-text

Assessing cost-effectiveness of early intervention in Alzheimer's disease: An open-source modeling framework

Alzheimer s & Dementia ◽

10.1016/j.jalz.2019.05.004 ◽

2019 ◽

Vol 15 (10) ◽

pp. 1309-1321 ◽

Cited By ~ 3

Author(s):

Colin Green ◽

Ron Handels ◽

Anders Gustavsson ◽

Anders Wimo ◽

Bengt Winblad ◽

...

Keyword(s):

Alzheimer’S Disease ◽

Alzheimer's Disease ◽

Early Intervention ◽

Cost Effectiveness ◽

Open Source ◽

Source Modeling ◽

Modeling Framework

Download Full-text

A non-parametric approach for co-analysis of multi-modal brain imaging data: Application to Alzheimer's disease

NeuroImage ◽

10.1016/j.neuroimage.2005.10.052 ◽

2006 ◽

Vol 30 (3) ◽

pp. 768-779 ◽

Cited By ~ 36

Author(s):

Satoru Hayasaka ◽

An-Tao Du ◽

Audrey Duarte ◽

John Kornak ◽

Geon-Ho Jahng ◽

...

Keyword(s):

Alzheimer’S Disease ◽

Alzheimer's Disease ◽

Brain Imaging ◽

Parametric Approach ◽

Imaging Data ◽

Data Application ◽

Brain Imaging Data ◽

Non Parametric

Download Full-text

Survival in Subcortical Vascular Dementia: Predictors and Comparison to Probable Alzheimer's Disease in a Tertiary Memory Clinic Population

Dementia and Geriatric Cognitive Disorders ◽

10.1159/000434626 ◽

2015 ◽

Vol 40 (3-4) ◽

pp. 210-221 ◽

Cited By ~ 4

Author(s):

Jong Hun Kim ◽

Seok Min Go ◽

Sang Won Seo ◽

Suk Hui Kim ◽

Juhee Chin ◽

...

Keyword(s):

Alzheimer’S Disease ◽

Alzheimer's Disease ◽

Family History ◽

Vascular Dementia ◽

Proportional Hazards Model ◽

Late Onset ◽

Cox Proportional Hazards ◽

Cox Proportional Hazards Model ◽

Subcortical Vascular Dementia ◽

History Of

Background: Subcortical vascular dementia (SVaD) is one of the most common dementias, after Alzheimer's disease (AD) dementia. Few survival analyses in SVaD patients have been reported. Methods: The dates and causes of death of 146 SVaD and 725 AD patients were included. We used the Cox proportional hazards model to compare survival between SVaD and AD patients and to explore possible factors related to survival of SVaD patients. Results: The median survival time after the onset of SVaD (109 months) was shorter than that recorded for AD (152 months). The most common cause of death in SVaD was stroke (47.1%). Factors associated with shorter survival in SVaD were late onset, male sex, worse baseline cognition, absence of hypertension and a family history of stroke. Conclusions: Stroke prevention may be important in SVaD treatment because 47.1% of SVaD patients died of stroke. A family history of stroke and absence of hypertension were associated with a shorter survival in SVaD, suggesting the existence of genetic or unknown risk factors.

Download Full-text