# Multi-timescale process models to disentangle subtle cognitive decline and learning effects

> **NIH NIH R56** · PENNSYLVANIA STATE UNIVERSITY, THE · 2021 · $668,993

## Abstract

Project Summary/Abstract
Sensitive and accurate repeated measurement of change in cognitive performance is necessary
for the detection of subtle cognitive decline in the preclinical phase of Alzheimer’s disease and
related dementias (ADRD). It is also required to evaluate outcomes of early interventions aimed
at mitigating advanced cognitive decline. However, this is a difficult task as these changes are
subtle not only in terms of magnitude, but also in terms of the latent processes through which
they manifest. In longitudinal studies researchers and clinicians are often interested in detecting
long-timescale (i.e., normative aging vs. disease progression) change patterns; however retest
related learning processes on short and long-timescales often confound these effects.
To address these challenges, we propose to develop a modern statistical toolset designed for
use with data from high-frequency repeated assessments over time. For this, we will combine
longitudinal measurement “burst” designs and a novel Bayesian statistical toolkit to capture
learning together with cognitive change and decline. These tools will provide interpretable
parameters (e.g., change in peak performance, probability of decline, caution in decision
making, etc.) that can then be deployed as digital biomarkers of subtle cognitive decline. The
Bayesian framework will also provide for a principled framework to communicate individual-
specific dementia risks towards clinicians. Our specific aims are to:
1. Disentangle cognitive change and decline, for example due to aging and/or disease
progression, from learning during repeated assessments over time, by implementing the multi-
timescale Bayesian double exponential learning model. This work can identify digital biomarkers
of cognitive decline associated with ADRD that are not confounded by learning effects.
2. Study individual differences in learning across multiple timescales and their links to normative
aging, and ADRD risks. This work can identify digital biomarkers of cognitive decline associated
with ADRD that articulated in terms of features of learning.
3. Delineate normative cognitive aging, ADRD risk, and features of learning by mapping them
onto latent cognitive sub-processes during task performance learning by developing the Drift
Diffusion Double Exponential Model framework. This work can identify digital biomarkers of
cognitive decline associated with ADRD that relate to sub-processes of cognitive performance
while also accounting for learning, as well as characterize sub-processes in which learning
occurs.

## Key facts

- **NIH application ID:** 10485503
- **Project number:** 1R56AG074208-01
- **Recipient organization:** PENNSYLVANIA STATE UNIVERSITY, THE
- **Principal Investigator:** Zita Oravecz
- **Activity code:** R56 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 2021
- **Award amount:** $668,993
- **Award type:** 1
- **Project period:** 2021-09-30 → 2023-08-31

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/10485503

## Citation

> US National Institutes of Health, RePORTER application 10485503, Multi-timescale process models to disentangle subtle cognitive decline and learning effects (1R56AG074208-01). Retrieved via AI Analytics 2026-05-24 from https://api.ai-analytics.org/grant/nih/10485503. Licensed CC0.

---

*[NIH grants dataset](/datasets/nih-grants) · CC0 1.0*