طراحي و ارزيابي يك سيستم تشخيص سقوط بلادرنگ سالمندان مبتني بر حسگر اينرسي

Design and eva‎luation of a real-time fall detection system for the elderly based on inertial sensor

مهندسي پزشكي

Detection and prevention of falls is important in geriatric health care, as it helps to reduce the harmful effects after a fall. In this context, a database, with the help of Unity game engine, has been developed that can replace public databases. To achieve this goal, we have developed an automatic feature extraction method that can extract temporal features from a variety of crash data. A deep neural classifier based on convolutional neural networks (CNN) and short-term memory networks (LSTMs) is trained on the extracted temporal features. The classifier was trained by the database made with Unity engine and achieved sensitivity, specificity and F1 score of 89 , 91 and 88% and Substantial agreement rate in SisFall database and 94, 87 and 90% and Substantial agreement rate in native database. This work can be used to design and develop fall detection devices for Internet-of-Healthcare-Things (IoHT) applications and to transfer fall detection capabilities to humanoid robots and gait rehabilitation devices such as exoskeleton robots and intelligent artificial legs.

5

77324