Sensors Fusion for Cognitive Load Analysis using Gait Data

ait is the manner of walking in people and one of the basic functions for humans to move purposefully to reach a desired destination. The quality of life can be affected by gait abnormality and result in morbidity and mortality. Substantially, our aim in this research is to develop new methods and algorithms that make the most of the existing sensors for gait analysis. A detailed review [1] reveals the existing achievements and gaps in the current knowledge in gait analysis. The modalities in literature to capture gait data are grouped according to the sensing technology: video sequences, wearable sensors, and floor sensors. Following from the review, sensors under the foot are identified as a suitable method to study gait deterioration due to cognitive load in this research. Therefore, Deep learning models are implemented to fuse sensors under the foot and deliver automatic feature extraction of gait patterns and perform classification for the following. (a) Gait under cognitive load difference in males and females [2], where both genders identified by 95% yet they share the same cognitive load by 93%. (b) Healthy subjects’ natural limits due to cognitive load capacity investigated using their gait. Layer-Wise Relevance Propagation technique is used to link key known events in the gait cycle to identify the influence of cognitive demanding tasks on gait [3]. (c) Parkinson’s disease staging based on postural imbalance caused by gait deterioration. The models classified patients’ gait by 96% using ground truth markers [4]. These findings present valuable insight for gait spatiotemporal signals analysis, with other potential spin-offs are in the areas of biometrics and security