Autonomous Human Monitoring and Activity Recognition for Safety and Health Improvement
TimeThursday, April 152:05pm - 2:06pm EDT
Human gait and activity monitoring play important roles in many applications that can have a significant positive impact on improving the ability to be as independent, secure, and healthy as possible. Notably, gait qualities are increasingly being recognized as a measure of a person’s health status; changes in normal (i.e., an individual’s usual) walking patterns, such as gait speed reduction or worsening balance, can signify a change in health and/or cognition. Variations in gait speed as a result of cognitive or other conditions may go undetected as the effect is often gradual and often not noticeable during clinic visits. Currently, gait measurements are conducted during medical visits, using subjective human-based estimates and/or expensive equipment, such as a mat-based system like GateRite. Infrequent measurements may not detect gradual changes, thus delaying or missing a diagnosis. This includes early identification of fall risk and the detection and swift response when falls occur (as falling is a leading cause of mortality and injuries among older adults). To support independence and quality of life as much as possible, both long-term detection of gradual changes in health, as well as the detection of acute events (e.g., a fall), is necessary. Most monitoring technologies currently rely on computer vision that employs cameras. While this approach has strengths, aspects that still need to be addressed include obstructed line-of-sight and privacy concerns. Wearable sensors / wearable technologies are also on the rise, however, these are not reliable as people are not always compliant (e.g., they forget or do not want to wear it), and the battery needs to be charged regularly for them to work. Wireless ambient sensors with the capability of sensing any environment dynamics without the need for wearable devices while preserving privacy are needed for robust monitoring of human health-related signals.
Develop a zero‐eﬀort and no-contact monitoring sensors to independently and autonomously recognize and instantly generate alerts for any detected fall and gait anomaly or other recognized ‘triggering’ event.
Radar-based systems, along with machine learning algorithms and artiﬁcial intelligence, are used to monitor gait and activities and to detect when and where a fall has occurred. The raw wireless signals will undergo extensive advanced signal processing, machine learning, and artiﬁcial intelligence algorithms. A continuous stream data will be generated and fed to the system, the wireless sensors coupled with AI then, given a new data point, will autonomously identify any deterioration and anomalies in an individual’s health, to recognize the level of activities, and to improve safety and health.
An ambient assisted living (AAL) system prototype for monitoring people (especially older adults) at hospitals, at the long-term care, and at home remotely without the people being monitored having to alter their activities in any way. A no-contact monitoring solution, delivering measurable improvements in quality of care through 24/7 always-on monitoring for the most vulnerable at-risk residents. The privacy-preserving technology using wireless sensors will provide a mechanism for the detection, prevention, and mitigation of health deterioration through the use of AI. The AAL system is a significant achievement in the development of autonomous continuous patient monitoring systems as it not only identifies anomalies and onset of issues immediately (e.g., falling down, change in gait patterns) but also has the potential to mitigate patient harm by predicting future health problem (e.g., dementia and mortality). The system is safe, reliable, and affordable, which could reduce health-care costs and support independence, security and safety for a wide variety of people who need caregivers and those who are at the risk of falls.