How Hard Am I Training? Using Smart Phones to Estimate Sport Activity Intensity

Smart phones are increasingly being used to track and recognize different types of activity. However, the task of using smart phones to infer the intensity of sport activities has not received a lot of attention yet. Therefore, we study how off-the-shelf smart phones with built-in accelerometers can be used to estimate the intensity of recreational sport activities. We focus on finding the most appropriate model along with a set of high level acceleration features that could be used to predict heart rate during a sport activity on a resource constrained smart phone device. We collect more than 300 minutes of acceleration and heart rate data from five subjects playing badminton and evaluate four different numeric prediction models using different combinations of acceleration features in terms of correlation between the actual and predicted heart rate and the heart rate estimation error. The evaluations show that linear regression provides good intensity inference accuracy (correlation coefficient: 0.86; mean absolute error: 15.52 beats per minute) and is, considering its low computational demands, the most feasible to be implemented on a smart phone device.