RESUMEN
Human falls rarely occur; however, detecting falls is very important from the health and safety perspective. Due to the rarity of falls, it is difficult to employ supervised classification techniques to detect them. Moreover, in these highly skewed situations, it is also difficult to extract domain-specific features to identify falls. In this paper, we present a novel framework, DeepFall, which formulates the fall detection problem as an anomaly detection problem. The DeepFall framework presents the novel use of deep spatio-temporal convolutional autoencoders to learn spatial and temporal features from normal activities using non-invasive sensing modalities. We also present a new anomaly scoring method that combines the reconstruction score of frames across a temporal window to detect unseen falls. We tested the DeepFall framework on three publicly available datasets collected through non-invasive sensing modalities, thermal camera and depth cameras, and show superior results in comparison with traditional autoencoder methods to identify unseen falls.
RESUMEN
People Living with Dementia (PLwD) often exhibit behavioral and psychological symptoms of dementia; with agitation being one of the most prevalent symptoms. Agitated behaviour in PLwD indicates distress and confusion and increases the risk to injury to both the patients and the caregivers. In this paper, we present the use of wearable devices to detect agitation in PLwD. We hypothesize that combining multi-modal sensor data can help in building better classifiers to identify agitation in PLwD in comparison to a single sensor. We present a unique study to collect motion and physiological data from PLwD. This multi-modal sensor data is subsequently used to build predictive models to detect agitation in PLwD. The results on Random Forest for 28 days of data from PLwD show a strong evidence to support our hypothesis and highlight the importance of using multi-modal sensor data for detecting agitation events amongst them.