Application of dynamic time warping algorithm for pattern similarity of gait

Area of Science:

• Biomechanics
• Gait Analysis
• Signal Processing

Background:

• Gait analysis is crucial in understanding human locomotion and identifying abnormalities.
• Traditional methods like the symmetry index (SI) have limitations in detecting subtle gait variations.
• Dynamic time warping (DTW) is a signal processing technique with potential applications in time-series data comparison.

Purpose of the Study:

• To evaluate the effectiveness of dynamic time warping (DTW) as a tool for gait similarity analysis.
• To compare the results of DTW with the commonly used symmetry index (SI) in gait research.

Main Methods:

• Collected three-dimensional (3D) motion data from 20 participants (10 males, 10 females) using an 8-camera infrared motion analysis system.
• Analyzed knee joint flexion angles using a custom MATLAB program implementing DTW.
• Compared DTW results with the symmetry index (SI) and performed statistical analysis using one-sample t-tests across various confidence intervals (CI).

Main Results:

• Dynamic time warping (DTW) yielded different results compared to the symmetry index (SI), suggesting DTW is more sensitive.
• The 10% threshold for similarity in SI was found to be too high compared to DTW's sensitivity.
• Using a 95% confidence interval, DTW showed a lower rate of classifying sequences as similar, even for normal gait.

Conclusions:

• Dynamic time warping (DTW) is a viable and potentially more accurate method for assessing gait similarity in research.
• DTW's sensitivity surpasses that of the traditional symmetry index (SI), offering finer discrimination of gait patterns.
• The findings support the use of DTW for rigorous gait similarity testing in scientific studies.