This growth of senior citizen’s number is due to improved living conditions, medical development and social progressthat lead to a worldwide increase in life expectancy. This unprecedented demographic change is a real challenge for society and the Swiss healthcare system, which must now face the price of this success: the increase in chronic pathologies affecting the elderly, the need of new retirement homes and the resulting costs. DOMO, a company from the École Polytechnique Fédérale de Lausanne (EPFL), designed an innovative system that uses machine learning to send personalized alarms in case of health deterioration or an emergency at home.
A system that promotes preventive medicine
The only solution with chronic pathologies is a 24h/24 health monitoring, which is normally very expensive andmtime consuming. DOMO has developed a system that easily detects a health deterioration by simply putting a bed sensor under the mattress. When the sensor detects a physical change (tachycardia, sleep apnea) it automatically sends an alert to the family or caregivers so that they can act preventively.
The entry into a retirement home: a moment to delay
In addition to chronic pathologies and costs increase, retirement homes will surely not be able to meet the growing demand. The objective of DOMO is to allow the elderly to stay independent at home as long as possible with a system that detects falls or unusual behavior and sends an alarm to the family or the emergency call center. This system is very useful for people suffering from dementia or Alzheimer because the network of wireless sensors (door, movement and bed sensors) detect if the person falls, goes outside during the night or doesn’t move during the entire day. A mobile app is also provided for relatives and caregivers that informs them about the person’s health, interacts with the caregivers and sets up personalized alarms according to the person’s lifestyle.
A system based on machine learning
During the first two weeks, DOMO’s technology records the lifestyle of the monitored person and establishes a behavior baseline. Then, if there is a change of behavior from the baseline, emergency or preventive alerts are shared with 24/7 call centers or home care organisations. The technology has been built with more than 200’000 of days analysed and uses unsupervised and supervised techniques. The clinical studies, conducted in collaboration with CHUV, the Inselspital of Bern and Nursing School of La Source in Lausanne, have been published in renown scientific newspapers such as Nature Scientific Report or Frontiers.