Every 19 minutes, an older adult dies from a fall
Dr. Stephen Robinovitch reports that 70% of falls were failed attempts at performing activities of daily living (ADL's). They are routine activities that people tend do every day without needing assistance. There are six basic ADLs: eating, bathing, dressing, toileting, transferring (walking) and continence.
Current Fall Detectors are not reliable
|
The current solution for detecting a fall is a wearable fall sensor (e.g. a pendant) or a smart watch. It measures body movements and position and attempts to differentiate between everyday movements and a fall. There are two problems. The first is remembering to wear the sensor and the second is the relatively poor performance in sensor accuracy. As summed up by Top Ten Reviews in a 2017 review (there has been no independent testing of smart watches)…
On average, devices detected about 75 percent of falls… While fall detection is a great addition to your medical alert system, you should not heavily rely on it but instead think of it as a backup. No fall detection sensor is or claims to be 100 percent accurate, and it is always best to push your alert button if able. |
Ne Reliable Technology
Senior Sentry is developing a fall detection system based upon artificial intelligence (deep learning) technology. Via wireless cameras, mounted on a wall or residing on a book shelf, data is collected 24/7/365 and falls are detected with 99% accuracy and the appropriate caregiver or responder notified.
The deep learning-based fall detection apps offer three distinct advantages. First, falls are detected regardless of the home environment. There can be confusing backgrounds, TV’s and pets. Secondly, no set-up is required. The fall detection algorithm is a generic solution (subject or person-independent), which means that it works out of the box and it does not need to do any subject-dependent training. And, lastly, over time the systems will learn - just as a human would.
The deep learning-based fall detection apps offer three distinct advantages. First, falls are detected regardless of the home environment. There can be confusing backgrounds, TV’s and pets. Secondly, no set-up is required. The fall detection algorithm is a generic solution (subject or person-independent), which means that it works out of the box and it does not need to do any subject-dependent training. And, lastly, over time the systems will learn - just as a human would.
