Categorizing Exercise Intensity Levels by Cardiorespiratory Signals

  • Dalia Y. Attia

Student thesis: Master's Thesis

Abstract

High-performance athletes are trying to get the most out of their workouts and achieve the maximal performance. However, training without knowing the limits can lead to overexertion and overtraining injuries. Therefore, the main goal of this study is to research for new methods to help athletes in controlling their exercise strategies and evaluating their exercise intensity levels. For this purpose, multiple simple non-invasive sensors were used to record electrocardiogram (ECG), respiration rate, body temperature, oxygen saturation level and body acceleration during exercise on different intensity levels from eight healthy male subjects. These recorded cardiorespiratory features and their Borg Rating of Perceived Exertion (PRE) scale associated with exercise intensity levels were used to train a multiclass Support Vector Machine (SVM) in order to predict exercise intensity levels during exercise. The developed model was used to classify data to three zones namely low, moderate and intense indicating exercise intensity levels. Moreover, the model was validated using several validation techniques and compared with other modeling techniques such as Naïve Bayes, K-nearest neighbor (KNN) and Trees. The model showed useful results for future research to differentiate between safe and danger zones for healthy and high-risk population to be able to perform exercise safely.
Date of AwardJun 2016
Original languageAmerican English

Keywords

  • Portable noninvasive sensors
  • Exercise intensity
  • Support Vector Machine
  • nonlinear multivariate model
  • Classifiers.

Cite this

'