UC Davis

Advances in communication technology and hardware performance have facilitated the widespread adoption of new technologies on smart devices. In return, this has led to greater availability, lower power consumption, and lower prices. With the swift progress of both artificial intelligence (AI) and the Internet of Things (IoT) technologies, the application of human activity recognition (HAR) has become increasingly widespread across a range of domains, including security and surveillance, human-robot interaction, entertainment, and healthcare. There is a trend of applying AI to HAR systems in order to achieve superior performance and accurate medical outcome results in real time. There are different approaches involving different implementation scenarios and challenges. An enormous number of proposals have been put forth to use AI to tackle these challenges while competing to be faster and with low power consumption. In this work, we focus on an AI-based solution for healthcare systems capable of identifying patients with muscular disorders, defining the degree of mobility limitation with a proper chart and scale, and identifying and monitoring characteristics that change over time with disease progression. Our work is a novel application of machine learning (ML) on mobile devices targeting rural healthcare. Our work demonstrates how the recent advancements in IoT devices and ML technology can be adapted to measure clinical outcomes, regardless of the point of care. This application can also be used for early clinical diagnosis and planning the course of treatment, as well as monitoring the disease progression.