iWatch: A Fall and Activity Recognition System Using Smart Devices

Abstract

Recent reports show that the average life expectancy is increasing worldwide, posing significant overhead on healthcare systems and increasing demands on long-term care facilities. One of the grand challenges directly related to growing ageing societies is the implications of falling. Many elderly people live alone, especially those in Western countries who cannot afford living in a senior house or retirement facility. In such cases, not only falling is a major concern, but also daily activities must be continuously monitored and analyzed to provide immediate support when needed. Vital signs and environment context are also crucial conditions for preand post-event assessments. Thanks to technology advancements and widespread adoption of the Internet of Things which enables us to provide smart and ubiquitous healthcare services. In this paper, we propose iWatch, a smart and flexible system for fall detection and activity recognition using common smart devices, a smartwatch and a smartphone. Machine learning techniques are used to build efficient and highly accurate activity recognition classifiers. iWatch also provides health risk analysis using threshold-based models and leverages visualization tools to better communicate with the user. iWatch is a promising technology that provides a small step in a giant leap to revolutionize healthcare services, especially for those who needs extra care.