Loading...

4,068 posters, 219 videos, 101 topics, 69 sessions, 14,365 authors, 6,128 institutions

ePostersLive by SciGen Technologies S.A. All rights reserved.

Alzheimer's Association International Conference 2026

July 12-15, 2026 | London, United Kingdom, and Online

Alzheimer's Association International Conference 2026 banner

7410

Functional Form Matters: How Model Specification Shapes Conclusions About Alzheimer's Disease Biomarker Relationships

Poster Presenter

Hunter Sylvester

Authors

Hunter Sylvester, Jeannette Simino, Murat Bilgel, Terry Therneau, Clifford Jack Jr., Petrice Cogswell, Ronald Petersen, David Knopman, Prashanthi Vemuri, Jon Graff-Radford, Thomas Mosley, Beverly Windham, Michael Griswold

Affiliations

Memory Impairment and Neurodegenerative Dementia (MIND) Center, University of Mississippi Medical Center, Jackson, Mississippi, USA, Laboratory of Behavioral Neuroscience, National Institute on Aging, Baltimore, Maryland, USA, Department of Radiology, Mayo Clinic-Rochester, Rochester, Minnesota, USA, Department of Neurology, Mayo Clinic-Rochester, Rochester, Minnesota, USA

Part of Topic

Title: 

Functional Form Matters: How Model Specification Shapes Conclusions About Alzheimer's Disease Biomarker Relationships

Background: 

•Alzheimer’s Disease (AD) pathology progression is often characterized by sequential S‑shaped shifts in biomarkers from normal to abnormal.
•Advanced modeling approaches are needed to capture nonlinear changes in multiple biomarkers simultaneously over time using longitudinal data.
•Key challenges in estimation include:
•Selecting optimal sigmoidal (S-shaped) functions for each biomarker
•Ensuring algorithmic convergence
•Producing interpretable and clinically meaningful estimates.  
 
Objectives: 
•We examine the sensitivity of AD cascade conclusions to functional form choice in Multivariate Nonlinear Mixed Effect Models (MNMs). We compare best-fitting, outcome-specific functional forms (Primary-MNM) selected using Watanabe-Akaike Information Criterion (WAIC) vs. assuming logistic functional forms across outcomes (Logistic-MNM). 
 
Contact Info: 

Hunter C. Sylvester hsylvester@umc.edu

The Memory Impairment and Neurodegenerative Dementia (MIND) Center

 

Methods: 

Study population: The Mayo Clinic Study of Aging (MCSA) is a prospective cohort from Olmsted County, Minnesota, ages 50 to 101 years (see Table 1).

Final samples: 

•1,576 PET amyloid SUVR measures
•495 plasma p-tau217 measures
•1,416 tau PET measures
•2,598 global cognition

Biomarkers: PET Amyloid SUVR (11C‑PiB PET, normalized to cerebellar gray matter, p-tau217 (C2N PrecivityAD2 LC‑MS assay), Tau PET, Global Cognition

Sigmoidal Functions:

•Logistic – Symmetric s-shaped1
•Hyperbola Function – Non s-shaped focusing on acceleration2
•Restricted Cubic Splines (RCS)

Statistical Analysis: MNMs with subject-specific age-shifts model longitudinal trajectories across age, adjusting for sex and APOE4 status. Estimates are presented as Bayesian posterior means with 95% credible intervals. 

 

 

Discussion: 

•Model choices can greatly affect results about the ordering of biomarker changes
•Best-Supported MNM supports biology more than assumed Logistic-MNM
•Cross-outcome correlations differed substantially between Primary-MNM and Logistic-MNM, particularly for tau PET with PET amyloid and plasma p-tau217

Limitations:

•~5% of participants had MCI/dementia, which may affect bend-point estimations

Comments

Email is required but is not shown publicly.

Loading comments...

aaic2026 banner
Copyright © 2015-2026, ePostersLive® SciGen is a registered trademark of SciGen Technologies S.A. Patent pending.

This website and third-party tools we use rely on cookies for the best user experience. By selecting "I agree", you agree to cookie usage as described in our Privacy Policy.