In order to improve the accuracy of fall detection,a personalized fall detection algorithm based on wearable devices and BP neural network is proposed.When the wearable device obtains the human movement data,it calculates the Ketole index according to the user's height and weight,constructs a feature dataset with a total of 1080 data pieces,classifies the dataset through BP neural network,and identifies the fall behavior.The test results show that the proposed algorithm has a recognition accuracy of 98.8％,a sensitivity of 97.9％,a specificity of 99.4％and a detection time of 0.27 s.Compared with the fall detec-tion algorithm that only takes the acceleration characteristic value as the detection data,the recognition accuracy of the pro-posed algorithm is increased by 4.9 percentage points,the sensitivity is increased by 2.9 percentage points,and the specificity is increased by 6.5 percentage points.This shows that the algorithm has high detection accuracy and real-time performance,which is suitable for the popularization of low-cost and high-performance wearable devices in the elderly group.
关键词
跌倒检测/可穿戴设备/BP神经网络/加速度传感器/克托莱指数
Key words
fall detection/wearable device/BP neural network/accelerometer transducer/Ketole index
维普
万方数据