Unfortunately, falls are very common among the elderly
population, which can lead to high health care costs and
mortality [1-3]. After a serious fall has occurred, the fear of
future falling can also tax a heavy psychological toll on the
patient, leading to functional declines and even eventually limit
the person to be placed in an institution for the elderly [4].
Numerous researchers have performed fall detection using
body-worn accelerometers [5-7]. Several accelerometer-based
fall detection reports use simple threshold values to
differentiate between falls and activities of daily living (ADL)
for patients. These systems mostly rely on a single threshold
value to detect whether an activity is considered as a fall or
simply an ADL, therefore are prone to false positives. In
addition, an accelerometer-only system cannot measure
minutes [3]. Such situations may have trapped the fallers in
danger no matter the falling is severe or not. The cost
forecasting of medical care for elderly residents’ fall-related
injury goes to $43.8 billion by 2020 [1]. No doubt it would be a
great burden to the national health insurance system.