Abstract

Purpose: Today, with the advancement of engineering sciences and technology in sports and physical health, the monitoring of physical activity has become the focus of many researchers. Accordingly, in this study, a new method has been proposed for classifying different sports activities based on the Photoplethysmogram signals. Methods: Four exercise activities, including running, walking, and cycling in light and heavy conditions, were used in this study. Data from the physionet database were used to perform the research, which included eight healthy participants with a mean age of 26.5. New criteria have been introduced for signal analysis based on the asymmetry of the delayed Poincaré’s plots. Results: The Wilcoxon statistical analysis results showed a significant difference between different sports activities (p < 0.05). A support vector machine was implemented for classification. To this effect, we have adopted a one vs all strategy. According to the classification results, the highest performance was related to walking detection, where the accuracy of 80.17%, the sensitivity of 81%, and the specificity of 75% were achieved. Conclusion: In summary, new indicators of delayed Poincaré’s plot can be effectively used to improve assessments of exercise activities or in designing personalized exercise programs.

Keywords

Photoplethysmogram, Delayed Poincare’s Plot, Asymmetry, Classification, Support Vector Machine