Fig 1.
Illustration of the data analysis workflow employed in this study.
Fig 2.
Summary of Silhouette values for k 2–5.
K2 and K3 resulted in identical data allocations for every repetition of k-means clustering. Due to a more variable cluster allocation for K4 the median cluster allocation is presented here. K5 cluster allocations displayed poor repeatability and a representative clustering outcome is presented.
Fig 3.
The effect of repeating k-means clustering ten times is presented with respect to the change in cluster allocation and the respective mean Silhouette value.
Across repetitions, data for two individuals changed cluster allocations once, three individuals changed twice, and one individual changed four times. As expected, the probability for individuals to change cluster allocations increases with decreased mean Silhouette values.
Fig 4.
Sagittal plane kinematics of the hip for participants allocated to clusters 1–4.
Mean data for participants with CP (red line) and normally developing children (green line) are shown (left vertical axis) for the duration of stance phase. Results from SPM analysis (black line) are shown with respect to the right axis. Shaded gray areas indicate significant differences between clusters and ND data. Abbreviations: Flexion (Flex), Extension (Ext).
Fig 5.
Coronal plane kinematics of the hip for participants allocated to clusters 1–4.
Mean data for participants with CP (red line) and normally developing children (green line) are shown (left vertical axis) for the duration of stance phase. Results from SPM analysis (black line) are shown with respect to the right axis. Shaded gray areas indicate significant differences between clusters and ND data. Abbreviations: Adduction (Add), Abduction (Abd).
Fig 6.
Transverse plane kinematics of the hip for participants allocated to clusters 1–4.
Mean data for participants with CP (red line) and normally developing children (green line) are shown (left vertical axis) for the duration of stance phase. Results from SPM analysis (black line) are shown with respect to the right axis. Shaded gray areas indicate significant differences between clusters and ND data. Abbreviations: Internal Rotation (Int), External Rotation (Ext).
Table 1.
Summary of mean kinematic pattern deviations for data allocated to each cluster with respect to a dataset of normally developing children.
The direction of observable pattern deviations and the timing of the deviations are indicated with respect to percent stance phase of walking.
Fig 7.
Knee sagittal plane kinematics for participants allocated to clusters 1–4.
Mean data for participants with CP (red line) and normally developing children (green line) are shown (left vertical axis) for the duration of stance phase. Results from SPM analysis (black line) are shown with respect to the right axis. Shaded gray areas indicate significant differences between clusters and ND data. Abbreviations: Flexion (Flex), Extension (Ext).
Fig 8.
Ankle sagittal plane kinematics for participants allocated to clusters 1-4.
Mean data for participants with CP (red line) and normally developing children (green line) are shown (left vertical axis) for the duration of stance phase. Results from SPM analysis (black line) are shown with respect to the right axis. Shaded gray areas indicate significant differences between clusters and ND data. Abbreviations: Dorsiflexion (Dorsi), Plantarflexion (Plant).
Table 2.
Summary of cluster characteristics [median and interquartile range (IQR)] with respect to the age, height, weight, and walking speed of individuals allocated to clusters 1–4.