Marcela Vallejo, Claudia V. Isaza, and Andrés Osorio
Geriatric Care, Fall Detection, Artificial Neural Network, Wearable Devices, Accelerometer
Falls are common events among elderly people and can have serious consequences. Opportune medical attention can minimize the repercussions of a fall and, for this reason, a large number of automatic fall detection devices have been proposed. The traditional approach to fall detection is based on the measurement of movement characteristics such as acceleration or velocity and the establishment of one or several thresholds. When these thresholds are exceeded, a fall is detected. The problem with threshold based systems is that some normal activities can also generate high acceleration and velocity, producing false alarms. On the other hand, if the selected threshold is a high value (in order to prevent false alarms), some falls can be undetected. In this paper a fall detection method based on artificial neural networks is presented which is able to differentiate between falls and normal activity while avoiding the use of thresholds. In addition, the experimental setup, necessary to obtain data to develop and test the algorithm, is described.
Important Links:
Go Back
