A method of posture monitoring and falling detection based on physiological and behavioral characteristics of the elderly

At present, the problem of population aging has become a hot spot of international concern, especially in China, and the international community urgently needs a universally applicable health care system for the elderly. Recent research shows that falling is the biggest threat to the health of the elderly. Based on thihe physiological and behavioral characteristics of the elderly, the paper discusses an algorithm for the recognition of motion state and fall detection of elderly applied to wearable devices to ensure timely rescue after a fall. The algorithm continuously acquires acceleration information during the movement of the elderly through a six-axis acceleration sensor. Firstly, the acceleration data is filtered, then the combined acceleration is calculated, and multiple features of the continuous data are extracted, and then the softmax method is used to classify the different motion states to realize the alarm of the fall. The algorithm extracts the feature vector by the magnitude of the combined acceleration, which solves the problem that the single acceleration in the traditional algorithm must solves the coordinate axis, which may waste much calculating time. The algorithm is validated by using the existing data set, and the accuracy of the algorithm is up to 89%. It is an effective way to detect falls.