An Algorithm for Fall Detection using Data from SmartWatch

Abstract

Falls are the leading cause of unintentional injuries in elderly people. These injuries need fast assistance in order to avoid severe consequences. In this context, some works have proposed approaches that use data from sensors such as accelerometer and gyroscope present in devices like smart-phones and smartwatches to detect falls as soon as they happen. However, current approaches still have deficiencies. Some of them need data from more than one device, which is not a typical use case. Other approaches could have better accuracy regarding activity recognition and fall detection. In this paper, we present an algorithm based on thresholds to detect falls that use information collected from a smartwatch accelerometer. To evaluate its performance, we compare it to two other threshold-based algorithms and the results indicate that our approach has a good accuracy in detecting falls and daily activities.