System evaluation for fall detection and prevention

Abstract

As the worlds population is becoming steadily older, the number of people in need of long-term care increases. With rising costs and the peoples wish for more autonomy, more devices supporting elderly people in everyday life are designed. Since falls leading to death are more likely to happen to the elderly, products enabling early help in falls or even preventing falls are coming into the market. This work compares chosen devices for fall prevention and fall detection. For the evaluation of fall prevention devices, the image-based sensor fearless, the motion sensor Optex and two versions of the pressure-based system Bucinator are used. After a 4-week field trial with 4 participants, 239 getup events are obtained and analysed. The assessment of devices used for fall detection is done in a laboratory setting comprising 10 subjects, simulating 11 fall scenarios and 11 Activities of Daily Living. The iWatch4 from Apple is compared with the fearless sensor, the mobile application b-cared and the accelerometer-based device sturzmelder. The precision and recall are calculated to evaluate the alarm behaviour of the fall prevention and fall detection devices. Speaking of fall prevention, fearless detects the most falls (92.93%), while it is 76.67% for the Bucinator Paulus. The mean false alarm rate of all detected alarms is for the Bucinator Paulus 15.00% and for fearless 32.47%. All of the tested devices used for fall detection send an alarm in under 30 seconds, which helps to minimize the health impact after a fall. With an adapted acceptance model, a questionnaire is performed to rise the user acceptance for AAL devices and in particularly for sensors detecting and preventing falls. 189 reply forms are evaluated, dividing the participants into the following groups: nursing staff, healthcare management, relatives and users aged 65 and older. 70.81% of all participants think that AAL technology will increase their job performance or facilitate things in everyday life. Image-based and non-image-based devices are analysed in terms of differences in the degree of privacy intrusion. Although researchers have made first steps towards a measurement of obtrusiveness, no metrics could be found to rate different devices. Audio recording as well as sending GPS, image and video data can be regarded as intrusive and is dependent on the users perception.