Abstract Details

Title Model-Based Truncation of Cognitive Test Batteries to Minimize Assessment Time and Information Loss

Topic Aging, Dementia, and Behavioral Neurology

Presentation(s) P1 - Poster Session 1 (5:30 PM-6:30 PM)

Poster/Presentation
Number 9-022

Objective

To build a minimal battery of cognitive tests that maximizes information about cognitive function while minimizing assessment time, given a theoretical model about cognition and empirical test data.

Background Cognitive testing is a critical component of standard neurological assessments in order to understand the nature and severity of cognitive impairment. However, developing a test battery that maximizes information about cognitive function while minimizing assessment time is not trivial. We used data from a recently introduced digital battery with nine well-known cognitive tests, Philips IntelliSpace Cognition (ISC), to demonstrate how a model of cognitive function can be used to make a selection of tests that balances assessment time and loss of clinically relevant information.

Design/Methods A latent variable model in line with literature on cognitive functioning was fit to the performance data of 148 healthy participants. Sensitivity to impairment was investigated for 48 TBI and 51 stroke patients. We developed an algorithm to iteratively generate smaller sets of tests taken from the ISC battery.

Results Our latent variable model for ISC mapped the outcomes of nine tests on six cognitive domains. TBI and stroke patients scored significantly lower than healthy participants on five cognitive domains. After applying our algorithm, we arrived at a smaller set of six tests, with only minimal information loss in domain score estimates and similar sensitivity to cognitive impairment in patient groups.

Conclusions We have shown that our algorithm, which operates on a theoretical model of test outcomes to cognitive domain mappings, is able to minimize the number of required tests while maintaining the same amount of information on cognitive functioning. These analyses showcase how an optimal set of tests can be selected from a battery in a model-based way.

Authors/Disclosures
PRESENTER	No disclosure on file
	No disclosure on file
	No disclosure on file
Justin B. Miller, PhD, ABPP/CN (Cleveland Clinic Foundation)	No disclosure on file
Ben Schmand, PhD	Dr. Schmand has received personal compensation in the range of $10,000-$49,999 for serving as a Consultant for Philips. Dr. Schmand has received publishing royalties from a publication relating to health care.

��ɫ��

��ɫ��