Dissemin is shutting down on January 1st, 2025

Published in

MDPI, International Journal of Environmental Research and Public Health, 17(19), p. 10977, 2022

DOI: 10.3390/ijerph191710977

Links

Tools

Export citation

Search in Google Scholar

Machine Learning for Prediction of Cognitive Health in Adults Using Sociodemographic, Neighbourhood Environmental, and Lifestyle Factors

This paper is made freely available by the publisher.
This paper is made freely available by the publisher.

Full text: Download

Green circle
Preprint: archiving allowed
Green circle
Postprint: archiving allowed
Green circle
Published version: archiving allowed
Data provided by SHERPA/RoMEO

Abstract

The environment we live in, and our lifestyle within this environment, can shape our cognitive health. We investigated whether sociodemographic, neighbourhood environment, and lifestyle variables can be used to predict cognitive health status in adults. Cross-sectional data from the AusDiab3 study, an Australian cohort study of adults (34–97 years) (n = 4141) was used. Cognitive function was measured using processing speed and memory tests, which were categorized into distinct classes using latent profile analysis. Sociodemographic variables, measures of the built and natural environment estimated using geographic information system data, and physical activity and sedentary behaviours were used as predictors. Machine learning was performed using gradient boosting machine, support vector machine, artificial neural network, and linear models. Sociodemographic variables predicted processing speed (r2 = 0.43) and memory (r2 = 0.20) with good accuracy. Lifestyle factors also accurately predicted processing speed (r2 = 0.29) but weakly predicted memory (r2 = 0.10). Neighbourhood and built environment factors were weak predictors of cognitive function. Sociodemographic (AUC = 0.84) and lifestyle (AUC = 0.78) factors also accurately classified cognitive classes. Sociodemographic and lifestyle variables can predict cognitive function in adults. Machine learning tools are useful for population-level assessment of cognitive health status via readily available and easy-to-collect data.