Human Psychophysiological Activity Detection Based on Wearable Electronics

Nowadays stress is an important problem of a society that is often the main reason for different diseases. For people, it is important to regularly monitor their health. However, it is not very convenient to go to medical centers or carry a lot of equipment every time. It is necessary to implement a means for monitoring health with the help of devices that we use in everyday life. Also, such devices must be able to analyze the human condition to exclude erroneous results and, as a result, erroneous influences. The paper considers modern approaches related to the human body state detection during various physical and mental activities. We describe the developed mobile application for collecting data on a person's state using connected wearable electronics. We identify physical and mental patterns of a person's state depending on various activities. We presented a classification model based on a random forest algorithm that can detect the patterns based on accelerometer data as well as heart rate measured by wearable electronics. We compared the results of classification accuracy using the random forest algorithm with other models based on logistic regression and AdaBoost.