Detecting fatigue in sit-to-stand transitions using knee flexion and dynamic time warping

Abstract

The sit-to-stand (STS) transition, a fundamental movement in daily life, is commonly used to assess strength, balance, and fall risk in older adults. Analyzing spinal kinematics during STS can provide valuable insights into physical health, particularly in individuals with spinal disorders or post-stroke. However, traditional methods for measuring these kinematics often involve expensive equipment and invasive procedures. This study, proposes a novel method for detecting fatigue in STS movements using video-based pose estimation without the need for attached sensors, focusing on knee flexion patterns and using dynamic time warping analysis. We used PoseNet to extract knee joints and analyzed knee angles using Dynamic Time Warping (DTW) to evaluate the consistency of movement before and after exercise. The results showed that Subject 5 had the lowest DTW value, reflecting more consistent and stable movement after exercise. In contrast, Subject 1 showed a significant difference with the highest DTW value, possibly due to fatigue. The results illustrated that the movement trajectory in the pre-exercise condition showed a smooth and controlled movement during the transition from sitting to standing. Meanwhile, in the post-exercise condition, there is a significant deviation in the knee trajectory, especially during the upward phase (standing). Observation of knee and hip trajectories in all subjects showed increased movement variability, especially in the knee after exercise, indicating that fatigue affects knee control more than the hip. This method successfully provided significant results regarding fatigue-related biomechanical changes without using expensive sensors. However, further research with a larger and more diverse number of subjects is needed.