Due to the influence of degeneration and
chronic diseases of elderly people, a higher chance of fall-related injuries
occurs among them. Falling is one of the accidents frequently confronted by
elderly people, so this issue is worthy of concern. We propose diverse models
to analyze falls through a wearable device. Then, we use Artificial
Intelligence of Things (AIoT) biomedical sensors for fall detection to build a
system for monitoring elderly people’s falls caused by dementia. The system can
meet the safety needs of elderly people by providing communication, position
tracking, fall detection, and pre-warning services. This device can be worn on
the waist of an elderly people. Moreover, the device can monitor whether or not
the person is walking normally, transmit the information to the rear-end
system, and inform his/her family member via a cellphone app while an accident
is occurring. Considering the risks on the fall test of elderly people, this
study adopts activities of daily living (ADL) to verify the test. According to
the test results, the accuracy of fall detection is 93.7%, the false positive
rate is 6.2%, and the false negative rate is 6.5%. To improve the accuracy of
fall detection and the timely handling of appropriate referrals, may be highly
expected to reduce the occurrence of fall-related injuries.
JEL classification numbers: D61, I30, O32.
Keywords: Fall Detection, AIoT Sensor, Elderly People.