SPECTRAL ANALYSIS OF DATA FROM A WRIST-WORN THREE-AXIS ACCELEROMETER TO DETERMINE PHYSICAL ACTIVITY

Abstract

Background. The relevance of the topic is determined by the increasingly widespread use of accelerometry data in the field of medicine, as well as in everyday life. The work raises the problem of automated determination of the type of human physical activity based on data collected using a wrist three-axis accelerometer. Materials and methods. To solve the problem, the method of analyzing the frequency spectrum of the signal was used. The accelerometer built into the smart watch was used as a source of initial data. To develop the data analysis program, the Scilab package of applied mathematical programs was used. Results. In order to analyze indicators of human physical activity, a program was developed that allows one to process and study data obtained from a three-axis accelerometer. For data analysis, a sliding window function has been implemented that shifts graphs at the touch of a button. Conclusions. A comparative analysis of graphs of frequency spectra and spectral differences showed that, along with the type of physical activity, the results are influenced by individual characteristics, which can complicate the identification of activities.

Keywords

• physical activity
• wearable accelerometer
• frequency spectrum
• sliding window
• scilab