College
College of Health Sciences
Department
Physical Therapy & Athletic Training
Program
Kinesiology and Rehabilitation
Publication Date
4-2021
DOI
10.25883/0n8t-7a88
Abstract
Lumbopelvic rhythm illustrates the relative motion between the lumbar spine and pelvis during various activities and could be used as a biomarker for low back pain (LBP). Sagittal plane lumbopelvic rhythm has been extensively examined as a surrogate to measure low back pain risk factor, but trunk rotation, the second component of lifting is commonly missed. Since lumbopelvic rhythm are time series and not discrete variables, machine learning may be a viable solution in identifying clusters of patterns for healthy adults. PURPOSE: To categorize healthy lumbopelvic rhythm in the transverse plane using machine learning. METHODS: 80 adults with no history of LBP (Young: n = 46; 26.9 ± 6.9 yr; Middle-Age: n = 33; 52.4 ± 6.9 yr). 3D kinematics of the lumbar spine and pelvis were calculated as participants performed maximal trunk rotation from right to left. Coupling angles were calculated using vector coding and represented in 4 coordination patterns (in-phase, anti-phase, superior-only, inferior-only). K-means clustering (k = 3) was used to segment coupling angles into clusters. Within each cluster, the age groups were compared. RESULTS: 3 distinct movement patterns were discovered (Figure 1). Lumbar spine and pelvis mostly moved in-phase, but for cluster 1, the start and end of the lumbar and pelvis was in anti-phase, while cluster 2 and 3 started and ended in-phase. Cluster 2 switched from in- to anti- and back to in-phase in the start and during transitioning directions. Age differences were seen only in cluster 1 where young and middle-age adults started rotation in anti-phase, but middle-age adults ended the rotation by only moving the lumbar spine and young adults ended in anti-phase. CONCLUSION: These movement patterns represent the different ways a healthy individual may perform trunk rotation, which along with sagittal plane motion can potentially be used to identify individuals with LBP.
Disciplines
Biomechanics | Kinesiology
Files
Download Full Text (399 KB)
Recommended Citation
Higgins, Seth; Tome, Joshua; and Kakar, Rumit Singh, "Using Machine Learning to Quantify Transverse Plane Lumbopelvic Rhythm" (2021). College of Health Sciences Posters. 8.
https://digitalcommons.odu.edu/gradposters2021_healthsciences/8