Elderly Image
An in-home study that collected data from non-contact sensors around elderly subjects’ residences used machine learning analysis to discover health problem indicators in human movement patterns. The research offers great potential for early detection of age-related issues, extending the age to which people safely live alone, and decreasing strains on the healthcare system.

An in-home study of the elderly that collected data from non-contact sensors around their residences combined with machine learning analysis has discovered health problem indicators in human movement patterns. The research from a team at the University of Bern and Bern University Hospital offers great potential for early detection of advanced age-related issues, extending the age to which people safely live alone, and decreasing strains on the healthcare system.

The team, led by Professor of Gerontechnology and Rehabilitation at the ARTORG Center for Biomedical Engineering Research Tobias Nef, used non-contact sensors placed in key areas throughout the 1,268 Swiss participants’ homes. These included motion sensors in each room, an under-mattress sensor, and sensors on doors to collect “an extensive collection of digital measures that capture broad parts of daily life, behavior, and physiology in order to identify [through specific changes in movement patterns] health risks of older people at an early stage.” These risks include detection of chronic diseases, such as dementia, Parkinson’s disease, and heart disease, that are often found late and are difficult to assess progression of, as well as fall risks, cognitive impairment, cardiac arrhythmia, COVID-19 complications, depression, and more.

When designing the study, the research team kept the subjects’ demographic in mind, thus, the choice to use non-contact, which the subjects favored as they disliked or found it physically difficult or impossible to use mobile devices daily. The sensors also recorded no sound or video and followed the highest levels of medical data security in Europe, avoiding another common concern of the demographic.

Sensor data on participants was fed into a program for machine learning analysis to identify possible digital biomarkers for the elderly. Nef explains, "For example, we found indications that fall risk could significantly depend on certain sleep parameters."

The research team expects the study enable older people to live alone until later in life, delaying entry into assisted living facilities and preventing unnecessary hospital stays.

Source