When elderly individuals experience falls at home, swift response is crucial, especially if they are alone.
A recent study conducted by Binghamton University, State University of New York introduces a novel approach to reduce response times using a human action recognition (HAR) algorithm. This algorithm leverages local computing power to analyze sensor data and identify unusual movements without the need to transmit data to an offsite processing center.
Professor Yu Chen and PhD student Han Sun, from the Department of Electrical and Computer Engineering at the Thomas J. Watson College of Engineering and Applied Science, developed the Rapid Response Elderly Safety Monitoring (RESAM) system. This system harnesses the latest advancements in edge computing.
Published in the IEEE Transactions on Neural Systems and Rehabilitation Engineering, their research demonstrates that the RESAM system can operate on common devices such as smartphones, smartwatches, laptops, or desktop computers with 99% accuracy and a rapid 1.22-second response time, making it one of the most precise methods available today.
Chen emphasized the significance of this research for an often overlooked demographic: “When many people discuss high tech, they usually refer to cutting-edge advancements. We identified a group of individuals – senior citizens – who require additional assistance but often lack the necessary resources or means to communicate their needs to developers of high-tech solutions.”
By utilizing devices already familiar to older adults, rather than requiring a complex “smart home” system, Chen believes that individuals feel more empowered in managing their health. They do not need to adapt to new technology for the system to be effective.
To maintain privacy, RESAM converts monitored images into simplified skeletons, enabling the analysis of essential points like arms, legs, and torso to detect falls or other incidents that may result in injuries.
“The bathroom is known to be one of the riskiest places for falls, but installing a camera there would be unwelcome,” explained Chen. “Nobody would approve of it.”
He envisions the RESAM system as a foundation for a broader concept he dubs “Happy Home,” which could incorporate thermal or infrared cameras and other sensors to remotely evaluate various aspects of an individual’s environment and well-being.
“Integrating additional sensors can enhance our system, as we are not only monitoring body movements but also monitoring health from another perspective, allowing us to predict potential incidents in advance,” he added.
An additional concept being explored by Chen and Associate Professor Shiqi Zhang from the Department of Computer Science involves integrating a robot dog or similar “pet” into the system. This virtual companion could offer closer supervision as individuals perform their daily routines. Zhang previously demonstrated how a robot dog could assist a visually impaired person by guiding them through leash tugs.
“Interaction with the robot is possible,” said Chen. “For instance, as you head to the bathroom, the dog might inquire, ‘May I accompany you?’ This dynamic approach allows the dog to adjust its position for closer monitoring instead of relying solely on fixed room sensors.”
