CNS Core: Small: Fundamentals of Gait Disorder Assessment with Ubiquitous Wireless Signals

University Of California-Santa Barbara, Santa Barbara CA

Investigators

Abstract

Neurological/brain-related gait disorders affect many people and can be devastating to the mobility, independence, cognition, and self-esteem of an individual. Thus, early diagnosis and proper monitoring is key to giving an individual the proper care that can optimize their well-being. Yet, many individuals seek a medical opinion only after the disease has progressed for a while. Furthermore, once medical help is sought, many patients miss follow-up appointments to monitor the progress of their disease under therapy/medication. The situation can be much worse in impoverished/developing nations (or rural areas in US), due to the high cost of healthcare and/or lack of it for an average family. Wireless signals, on the other hand, are ubiquitous these days, which open up the possibility of using them for sensing and learning about the environment. This is the main motivation for the proposed work, to introduce a new mathematical foundation and design methodology that enables everyday RF signals, such as WiFi, to detect, classify, and monitor gait disorders. The proposed work can result in a considerable advancement towards developing an affordable home health system for gait disorder assessment. Such a system can further work in partnership with medical professionals, for the diagnosis and monitoring of gait disorders. The project also has an educational component, targeting K-12 as well as under-represented groups. This research proposes a new foundation for gait disorder assessment with ubiquitous RF signals. This is a considerably challenging problem, which is divided into four major tasks. The first major goal proposes a new methodology that can translate the vast already-available online non-RF gait disorder datasets to RF data, enabling the creation of a large synthetic gait disorder RF datasets pertaining to different gait disorders. Such datasets are necessary for a methodical analysis/design, but currently lacking and prohibitively laborious to manually collect. The second major task then proposes a new processing foundation that can mathematically characterize, for the first time, the general full frequency content of the received signal, and its high-energy epochs, enabling proper analysis of more complex movements. The third objective then builds on the previous two to extract rich, efficient, and meaningful features from the received signal and design a robust machine leaning pipeline for the detection, classification, and monitoring of gait disorders, with an emphasis on understanding the feasibility and limitations. Finally, the last objective validates the proposed foundation with extensive experiments. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.