INS NYC 2024 Program

Poster	Poster Session 06 Program Schedule 02/15/2024 04:00 pm - 05:15 pm Room: Shubert Complex (Posters 1-60) Poster Session 06: Aging \| MCI \| Neurodegenerative Disease - PART 2 Final Abstract #39 Modeled Location Learning Test-Revised learning curves to differentiate minor and major neurocognitive disorders Willem Eikelboom, Vincent van Gogh Institute for Psychiatry, Centre of Excellence for Korsakoff and Alcohol-Related Cognitive Disorders, Venray, Netherlands William Goette, University of Texas Southwestern Medical Center, Department of Psychiatry, Texas, United States Yvonne Rensen, Vincent van Gogh Institute for Psychiatry, Centre of Excellence for Korsakoff and Alcohol-Related Cognitive Disorders, Venray, Netherlands Gwenny Janssen, Vincent van Gogh Institute for Psychiatry, Centre of Excellence for Korsakoff and Alcohol-Related Cognitive Disorders, Venray, Netherlands Roy Kessels, Vincent van Gogh Institute for Psychiatry, Centre of Excellence for Korsakoff and Alcohol-Related Cognitive Disorders, Venray, Netherlands Category: Memory Functions/Amnesia Keyword 1: learning Keyword 2: memory disorders Keyword 3: neuropsychological assessment Objective: Episodic memory learning curves represent the ability to acquire new information across repeated trails of a memory test. Learning curves of episodic verbal memory tests are shown to provide valuable diagnostic information on the process of memory acquisition and can differentiate minor from major neurocognitive disorders (NCD). Spatial memory is a prominent aspect of episodic memory and spatial memory tests are able to capture subtle cognitive deficits and are strongly related to everyday functioning. Yet, studies on learning curves of spatial memory tests are limited. Therefore, we compared spatial memory learning curves between various NCD and investigated whether spatial memory learning may differentiate minor NCD from major NCD within patient groups. To do so, we used the Location Learning Test-Revised (LLT-R), which is a visuospatial analogue of verbal word-learning tests. Participants and Methods: We included a total of n=1,012 individuals across two patient groups including Alzheimer’s disease (AD) (minor NCD N=139, major NCD N=129) and vascular cognitive impairment (VCI) (minor NCD N=64, major NCD N=33), together with N=647 healthy controls (HC). All individuals completed the LLT-R. They were shown 10 pictures of everyday objects on a 5x5 grid, after which the objects had to be relocated on an empty grid. This procedure was repeated for five learning trials, where increased learning performance is indicated by fewer displacement errors. We fitted learning curves over these five immediate recall trials using Bayesian generalized non-linear mixed regression. Modeling was based on three parameters: initial memory performance (Attention), minimum number of displacement errors (Maximum Learning), and the decrease of displacement errors over the trails (Learning Rate). Results: An intercept-only zero-inflated negative binomial model demonstrated good fit to the data with an adjusted R² of 0.80 (95% CI: 0.79-0.81) and intraclass correlation of 0.73 (95% CI: 0.71-0.76). Bayesian ANOVA showed very strong evidence (BF₁₀≫100) for differences in Attention between all groups (AD<VCI<HC). Similarly, very strong evidence was found for differences in Learning Rate (AD<VCI<HC). In contrast, anecdotal to moderate evidence (0.93<BF₀₁<6.42) was found for the equivalence of Maximum Learning between the groups. Utilizing a holdout sample of N=100 participants, the LLT-R learning curves predicted from the model were able to correctly classify 56% of etiologies and severities while the sum score across learning trials classified no more than 51%. The model correctly classified 79% of HC, 68% of AD, and 13% of VCI cases in the holdout sample, while sum score and demographics correctly classified 75% of HC, 29% of AD, and 11% of VCI. Conclusions: Modeled spatial memory learning curves parameters Attention and Learning Rate were able to differentiate NCD etiologies and severities, while this was not the case for Maximum Learning. Classification abilities of the model was similar compared to traditional spatial memory measures, although the model performed better for individuals with AD. Modeled spatial memory learning curves appears a valid learning index in NCD. Analyses are ongoing and will be extend to N=225 individuals with minor or major alcohol-related cognitive disorders.

Poster

02/15/2024
04:00 pm - 05:15 pm
Room: Shubert Complex (Posters 1-60)

Poster Session 06: Aging | MCI | Neurodegenerative Disease - PART 2

Final Abstract #39

Modeled Location Learning Test-Revised learning curves to differentiate minor and major neurocognitive disorders

Willem Eikelboom, Vincent van Gogh Institute for Psychiatry, Centre of Excellence for Korsakoff and Alcohol-Related Cognitive Disorders, Venray, Netherlands
William Goette, University of Texas Southwestern Medical Center, Department of Psychiatry, Texas, United States
Yvonne Rensen, Vincent van Gogh Institute for Psychiatry, Centre of Excellence for Korsakoff and Alcohol-Related Cognitive Disorders, Venray, Netherlands
Gwenny Janssen, Vincent van Gogh Institute for Psychiatry, Centre of Excellence for Korsakoff and Alcohol-Related Cognitive Disorders, Venray, Netherlands
Roy Kessels, Vincent van Gogh Institute for Psychiatry, Centre of Excellence for Korsakoff and Alcohol-Related Cognitive Disorders, Venray, Netherlands

Category: Memory Functions/Amnesia

Keyword 1: learning
Keyword 2: memory disorders
Keyword 3: neuropsychological assessment

Objective:

Episodic memory learning curves represent the ability to acquire new information across repeated trails of a memory test. Learning curves of episodic verbal memory tests are shown to provide valuable diagnostic information on the process of memory acquisition and can differentiate minor from major neurocognitive disorders (NCD). Spatial memory is a prominent aspect of episodic memory and spatial memory tests are able to capture subtle cognitive deficits and are strongly related to everyday functioning. Yet, studies on learning curves of spatial memory tests are limited. Therefore, we compared spatial memory learning curves between various NCD and investigated whether spatial memory learning may differentiate minor NCD from major NCD within patient groups. To do so, we used the Location Learning Test-Revised (LLT-R), which is a visuospatial analogue of verbal word-learning tests.

Participants and Methods:

We included a total of n=1,012 individuals across two patient groups including Alzheimer’s disease (AD) (minor NCD N=139, major NCD N=129) and vascular cognitive impairment (VCI) (minor NCD N=64, major NCD N=33), together with N=647 healthy controls (HC). All individuals completed the LLT-R. They were shown 10 pictures of everyday objects on a 5x5 grid, after which the objects had to be relocated on an empty grid. This procedure was repeated for five learning trials, where increased learning performance is indicated by fewer displacement errors. We fitted learning curves over these five immediate recall trials using Bayesian generalized non-linear mixed regression. Modeling was based on three parameters: initial memory performance (Attention), minimum number of displacement errors (Maximum Learning), and the decrease of displacement errors over the trails (Learning Rate).

Results:

An intercept-only zero-inflated negative binomial model demonstrated good fit to the data with an adjusted R² of 0.80 (95% CI: 0.79-0.81) and intraclass correlation of 0.73 (95% CI: 0.71-0.76). Bayesian ANOVA showed very strong evidence (BF₁₀≫100) for differences in Attention between all groups (AD<VCI<HC). Similarly, very strong evidence was found for differences in Learning Rate (AD<VCI<HC). In contrast, anecdotal to moderate evidence (0.93<BF₀₁<6.42) was found for the equivalence of Maximum Learning between the groups. Utilizing a holdout sample of N=100 participants, the LLT-R learning curves predicted from the model were able to correctly classify 56% of etiologies and severities while the sum score across learning trials classified no more than 51%. The model correctly classified 79% of HC, 68% of AD, and 13% of VCI cases in the holdout sample, while sum score and demographics correctly classified 75% of HC, 29% of AD, and 11% of VCI.

Conclusions:

Modeled spatial memory learning curves parameters Attention and Learning Rate were able to differentiate NCD etiologies and severities, while this was not the case for Maximum Learning. Classification abilities of the model was similar compared to traditional spatial memory measures, although the model performed better for individuals with AD. Modeled spatial memory learning curves appears a valid learning index in NCD. Analyses are ongoing and will be extend to N=225 individuals with minor or major alcohol-related cognitive disorders.