Development and comparison of customized voice-assistant systems for independent living older adults

Abstract

Voice-controlled in-home personal assistants have great potential to assist older adults. This thesis explores the aspects of human-computer interface design, specifically a voice assistant, to help older adults manage their personal health, especially in the case of chronic health conditions. In our previous work, we have built a web interface for caregivers to monitor older adults' health changes based on in-home sensor data from motion sensors, bed sensors, and depth sensors. Data collected from these sensors are stored in servers and processed using several algorithms to obtain health and activity parameters including gait, fall risk, detect fall, motion patterns, sleep, heart rate, and respiration rate, as well as to generate health alerts. The sensor system with automated health alerts and care coordination has been shown to help seniors maintain better functionality. In our current research project, we focus on developing a consumer interface for older adults and their designated trusted others that can provide health information on-demand, based on spoken queries. The health information is presented as both audio and visual displays and uses graphical visualizations and linguistic summaries of the sensor data trends and changes. The goal is to present data in a form that is simple to understand. To accomplish our objective of creating an easy-to-use-and-understand health data interface for older adults, we explore voice-controlled, in-home personal assistants as a solution. Two voice assistant platforms with displays were selected for implementation and testing, namely, the Amazon Echo Show and the Lenovo Smart Display with Google Assistant.