SDQI ‑Fall Detection System for Elderly

Abstract

Falls suffered by the elderly represent a serious risk for this part of the population, a risk that has attracted attention of geriatric in recent years. One of the main problems is the lack of rescue after the falls occurrence, which can extends for minutes, hours or even days. These long periods without help can cause irreversible sequels to the elderly lives. This paper proposes a solution for automated detection of these falls, using different approaches (detection of inactivity, detection of falls by thresholds analysis, detection of falls by device orientation analysis and detection of falls with decision trees algorithm) together, in order, to improve the efficiency and accuracy of the fall detection process. The data analyzed by the detector are provided by tri-axial accelerometers, which in this study were provided by smartphones with Android operating system. Through the databases Mobifall, Mobifall2 and a database developed by this study, tests performed with the proposed methodology showed 87.65% of specificity and 95.45% of sensitivity, with maximum detection delay of 3 seconds.