Zero-effort ambient vitals monitoring

点击翻译按钮获取中文摘要

Vital signs (e.g. respiration, blood pressure, heart rate) provide a wealth of information about a person's health and their frequent measurement is fundamental to diagnosing, monitoring, and managing conditions such as congestive heart failure, kidney failure, and coronary artery disease. Remote monitoring at home with traditional devices (e.g. blood pressure cuff) and wearable devices (e.g. smart watch) can provide information crucial to disease management and treatment, which can improve outcomes, increase independence, and reduce visits to a clinic or hospital. However, these technologies require training and diligent use to be effective, which is not a viable option for people who are not able to use these technologies reliably, such as people with dementia, people with arthritis and other mobility impairments, and overburdened care partners. This means that current remote monitoring methods are excluding a large portion of the population who stand to benefit significantly, including many older adults. We need new technologies that can reliably sample vitals in a home environment without any effort from the people using them.******The purpose of this program is to investigate zero-effort vital signs monitoring approaches. We will develop sensors, data processing, and machine learning techniques to create a cushion-like device that can be placed on any chair or sofa to autonomously capture vitals signs while a person is sitting on it. As conventional monitoring requires skin contact, a key research challenge is to derive a combination of sensors and signal processing methods that can acquire vitals through clothing. This will be accomplished by fusing existing sensors with novel carbon nanotube ones. We will build on signal processing methods such as adaptive and non-linear particle filtering to remove noise caused by movement and isolate signals of interest.******This research will also investigate the use of sensors such as thermal cameras and high-frequency radar to capture a person's activity prior to sitting down. We will apply machine learning to autonomously calibrate the device to different people and connect vitals to their level of activity. Correlating changes in vital signs (e.g. elevated heart rate) captured by the cushion to data from the activity sensors about what a person was doing before the measurement was taken is a novel approach that will enable a more comprehensive representation of a person's health.******The knowledge and technology resulting from this research represents pivotal advancements in sensors and machine learning that will enable data collection in non-clinical settings with associated context for a more holistic understanding of a person's health. This research will advance Canada's position as a leader in ambient monitoring technology and will benefit many application areas, including providing much-needed support for our globally aging population.