Combining wearable device with machine learning for intelligent health detection

The aging of the population gradually intensifies. The health status and medical problems of the elderly have aroused widespread concern in society. Wearable health monitoring system is a typical application of wearable computing in the medical field, which can achieve continuous and dynamic acquisition of human status under low physiological and psychological loads. Fall detection monitoring plays an important role in eldercare. This paper establishes a wearable health monitoring system for fall detection based on a three-axis accelerometer. First, the acceleration signals are collected through a three-axis accelerometer which is installed into a wearable device. Second, the collected acceleration signals are represented as 20 features, including mean of acceleration signal, SD of acceleration signal, coefficient Kurtosis, coefficient of skewness etc. Third, the acceleration signal features are used to learn a covariance-guided one-class support vector machine due to the difficulty to obtain fall acceleration signals. The experiments and simulations show the effectiveness of the proposed system for fall detection.