Multimodel Lightweight Transformer Framework for Human Activity Recognition

Human Activity Recognition (HAR) finds extensive application across diverse domains. Yet, its integration into healthcare remains challenging due to disparities between prevailing HAR systems optimized for rudimentary actions in controlled settings and the nuanced behaviors and dynamic conditions pertinent to medical diagnostics. Furthermore, prevailing sensor technologies and deployment scenarios present formidable hurdles regarding wearability and adaptability to heterogeneous environments. While navigating these constraints, this investigation evaluates the requisite monitoring simplicity and system adaptability crucial for medical contexts. A HAR framework is proposed, leveraging a Lightweight Transformer architecture with a multi-sensor fusion strategy employing five Inertial Measurement Units (IMUs) as sensors. A Real-world HAR dataset is assembled to authenticate the system’s suitability, and a comprehensive array of experiments is conducted to showcase its potential utility.

I. Introduction

Human Activity Recognition (HAR) is an invaluable technological asset employed across diverse domains, demonstrating notable significance within the biomedical realm. HAR plays a pivotal role in this domain across diverse areas, encompassing sports rehabilitation medicine (SRM), geriatrics, and neurology. In SRM, HAR facilitates the continuous monitoring of stroke patients’ daily activities, providing practitioners with the means to evaluate the effectiveness of rehabilitation interventions [1]. Furthermore, HAR harnesses activity levels and behavioral patterns to extract insights into the health status of elderly individuals, thereby enabling the timely identification and management of potential health concerns or safety risks they may encounter [2]. Within neurology, HAR is crucial in monitoring patients afflicted with neurological conditions such as Parkinson’s, detecting movement disorders, and offering an objective assessment of their progression [3], [4]. In summary, HAR’s ability to analyze diverse human body data elevates its utility across various biomedical applications, rendering it an indispensable tool in clinical practice and research endeavors.