Evaluation of stride variability between experienced and novice runners during a prolonged run

The objectives of this dissertation were to evaluate the accuracy of using inertial measurement unit (IMU) for gait events prediction during overground running and investigate effects of running experience and fatigue on stride-to-stride variability during a prolonged treadmill run. The entire study consisted of three parts. The first part evaluated the accuracy of three typical IMU-based methods for gait events prediction during an overground run. The S-method, placing the IMU at the shoe instep and analysing the resultant acceleration, produced the most accurate initial contact prediction with mean absolute difference of 4.7 (4.1) ms. The M-method, placing the IMU at the shank and analysing the vertical acceleration, produced the most accurate toe-off prediction with mean absolute difference of 7.0 (3.5) ms. The MS-method—a combination of the S- and M-methods—provided the most accurate stance time estimation with mean percentage difference of 3.8% (1.6%), and mean absolute differences of 9.1 (4.2) ms during jogging and 8.8 (3.5) ms during running.
The second part investigated stride interval dynamics of both experienced and novice runners while performing a 31-min treadmill run at their individual anaerobic threshold speeds. The scaling exponent alpha of the detrended fluctuation analysis and coefficient of variance were used to quantify the complexity and variability of the stride interval dynamics, respectively. A U-shape trend of the alpha was observed for both the experienced and novice runners, but the two groups presented slight differences in both the alpha and the coefficient of variance. Both the experienced and novice runners regulated the stride interval complexity to maintain the run at anaerobic threshold speed. The experienced runners also regulated the stride interval variability.
The final part investigated lower-limb coordination pattern and variability during a 31-min treadmill run. Lower-limb coordination pattern and variability during the stance phase at the beginning, middle, and end of the run were quantified using a modified vector coding technique. Running experience and progressive fatigue had significant interactions on the coordination patterns for the hip–knee and pelvis–thigh couplings. The experienced runners exhibited a higher percentage of in-phase motion for the pelvis–thigh and knee–ankle couplings, whereas the novice runners exhibited a higher percentage of distal dominant motion for the pelvis–thigh coupling and anti-phase motion for the hip–knee coupling during mid-stance. The experienced runners exhibited more variability in the hip–knee and shank–foot couplings, whereas the novice runners had more variability in hip, knee, and thigh motions. The experienced and novice runners adapted to fatigue through different lower-limb coordination patterns. The experienced runners demonstrated larger variability for segment/joint coupling motions and the novice runners exhibited larger variability for single segment/joint motions.
In conclusion, this dissertation demonstrated that initial contact and toe-off could be most accurately predicted by identifying the local peak resultant acceleration measured by the foot IMU and the minimum vertical acceleration measured by the shank IMU during overground running, and different gait regulation strategies for adapting to progressive fatigue between the experienced and novice runners during treadmill running at anaerobic threshold speed. All rights reserved.