Change detection and convolution neural networks for fall recognition

Accurate fall detection is a crucial research challenge since the time delay from fall to first aid is a key factor that determines the consequences of a fall. Wearable sensors allow a reliable way for motion tracking, allowing immediate detection of high-risk falls via a machine learning framework. Toward this direction, accelerometer devices are widely used for the assessment of fall risk. Although there exist a plethora of studies under this perspective, several challenges still remain, such as dealing simultaneously with extremely demanding data management, power consumption and prediction accuracy. In this work, we propose a complete methodology based on the cooperation of deep learning for signal classification along with a lightweight control chart method for change detection. Our basic assumption is that it is possible to control computational resources by selectively allowing the operation of a relatively heavyweight, but very efficient classifier, when it is truly required. The proposed methodology was applied to real experimental data providing the reliable results that justify the original hypothesis.