In-lab versus at-home activity recognition in ambulatory subjects with incomplete spinal cord injury

This study explores the accuracy of activity recognition algorithms for individuals with incomplete spinal cord injury, comparing in-lab and at-home data. The research highlights that algorithms trained on in-lab data may not accurately classify activities performed at home due to differences in movement patterns. Tailoring activity recognition algorithms using at-home data can significantly improve the accuracy of tracking movements in real-world settings for this population.

• 1
  Classifiers trained and tested using within-subject cross-validation in the lab provided an accuracy of 91.6%.
• 2
  When the classifier was trained on data collected in the lab but tested on at home data, the accuracy fell to 54.6% indicating distinct movement patterns between locations.
• 3
  The accuracy of the at-home classifications, when training the classifier with at-home data, improved to 85.9%.