Use of Machine Learning to Establish the Role of Cognitive Stress in Postural Sway Based on Multimodal Physiological Features

Abstract

Maintaining balance through postural control is crucial for everyday activities like sitting, standing, and reacting to balance challenges. Despite these tasks being largely automatic, they still require some degree of cognitive processing and adaptive processes. Cognitive stress can impact postural control, as shown by previous studies demonstrating its effect on postural sway. While previous research has shown that cognitive stress can affect postural sway, more indepth exploration is needed to better understand this relationship using advanced machine learning techniques and multimodal physiological features to gain a better understanding of the relationship between cognitive function, postural control, and physiological processing.

This master’s thesis uses multimodal physiological features to study the impact of cognitive stress induced by Stroop tests (English, Arabic, and mixed) on postural sway. Using machine learning models following data extraction, the study aims to investigate the underlying effects of language control (Stroop) cognitive stress on postural sway among young adults. Postural sway will be assessed through IMU recordings, and the physiological signals will be recorded using the NeXus-10 while standing with eyes open. The classification models were used to classify high and low postural sway using the features extracted from the recordings. The best performing models were logistic regression and support vector machine.

With a focus on the young adult population in the United Arab Emirates, the dataset generated could serve as a foundation for expanding research to other regions, offering potential preventive measures for serious diseases. In summary, this master’s thesis uses machine learning models to explore the relationship between cognitive stress, multimodal physiological features, and postural sway to better understand the impact of cognitive stress on posture through multimodal physiological measures.

Date of Award	19 Jul 2024
Original language	American English
Supervisor	Jelinek (Supervisor)