Related Experiment Video
Updated: Jun 5, 2025
12:18
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
7.5K
A multi-stage feature selection method to improve classification of potential super-agers and cognitive decliners
Parvin Mohammadiarvejeh1, Mohammad Fili2, Alice Dawson3
1University of Maryland Medical System, Linthicum, MD, USA.
Geroscience
|December 10, 2024
Summary
Older adults maintaining high cognitive function show distinct brain structures compared to those with cognitive decline. This research identifies key brain regions and factors associated with positive cognitive aging, potentially aiding in Alzheimer's disease risk reduction.
Area of Science:
- Neuroscience
- Gerontology
- Cognitive Science
Background:
- Cognitive aging involves age-related declines in memory, executive function, and processing speed.
- Super-Agers (over 80) exhibit cognitive performance comparable to middle-aged adults.
- Understanding brain differences between positive and declining cognitive aging is crucial for cognitive reserve and Alzheimer's disease (AD) risk reduction.
Purpose of the Study:
- To contrast regional brain volumes between "Positive-Agers" and "Cognitive Decliners" to enhance understanding of cognitive reserve.
- To identify key predictors of cognitive trajectory using structural magnetic resonance imaging (sMRI) and demographic data.
Main Methods:
- Longitudinal cognitive tests (7-9 years) from UK Biobank were used to derive a general cognition (GC) composite score via principal component analysis (PCA).
- A multi-stage feature selection algorithm identified important sMRI and covariate predictors for cognitive groups.
- A Random Forest (RF) classifier was trained on 54 selected features to differentiate between Positive-Agers and Cognitive Decliners.
Main Results:
- The RF model achieved an Area Under the Curve (AUC) of 73% in distinguishing cognitive groups.
- Key predictors included age, education, total brain surface area, pars orbitalis area, thalamus mean intensity, and superior frontal gyrus surface area.
Conclusions:
- Structural brain imaging can predict cognitive trajectory types.
- Identifying factors associated with successful cognitive aging may offer insights into reducing AD risk and improving cognitive reserve.
