TY  - JOUR
PY  - 2017//
TI  - Selecting power-efficient signal features for a low-power fall detector
JO  - IEEE transactions on bio-medical engineering
A1  - Wang, Changhong
A1  - Redmond, Stephen
A1  - Lu, Wei
A1  - Stevens, Michael
A1  - Lord, Stephen
A1  - Lovell, Nigel
SP  - 2729
EP  - 2736
VL  - 64
IS  - 11
N2  - Falls are a serious threat to the health of older people. A wearable fall detector can automatically detect the occurrence of a fall and alert a caregiver or an emergency response service so they may deliver immediate assistance, improving the chances of recovering from fall-related injuries. One constraint of such a wearable technology is its limited battery life. Thus, minimization of power consumption is an important design concern, all the while maintaining satisfactory accuracy of the fall detection algorithms implemented on the wearable device. This paper proposes an approach for selecting power-efficient signal features such that the minimum desirable fall detection accuracy is assured. Using data collected in simulated falls, simulated activities of daily living, and real free-living trials, all using young volunteers, the proposed approach selects three features from a set of ten commonly-used features, providing a power saving of 75.3%, while limiting the error rate of a binary classification decision tree fall detection algorithm to 7.1%.<p />  <p>Language: en</p>
LA  - en
SN  - 0018-9294
UR  - http://dx.doi.org/10.1109/TBME.2017.2669338
ID  - ref1
ER  -