Clinical Sciences/Health Conditions
Shintaro Fujii, PhD
Physical Therapist
Nishiyamato Rehabilitation Hospital
Kitakatsuragi-gun, Nara, Japan
Kenta Takeda, PhD
Assistant Professor
Japan Healthcare University
Sapporo, Hokkaido, Japan
Kosuke Sakano, BS
Physical Therapist
Hokkaido Neurological Hospital
Sapporo, Hokkaido, Japan
Noritaka Kawashima, PhD
Section Chief
Research Institute of National Rehabilitation Center for Persons with Disabilities
Tokorozawa, Saitama, Japan
Quiet-standing postural sway is commonly quantified using center of pressure (CoP) metrics; however, Parkinson’s disease (PD) involves additional postural abnormalities such as stooped posture and tibialis anterior (TA) overactivity that are not captured by CoP alone. A multidimensional approach integrating biomechanical and neuromuscular signals is therefore required to delineate the mechanisms underlying postural instability in PD. This study aimed to identify latent components of postural control and classify PD-specific postural phenotypes using exploratory factor analysis (EFA) and cluster analysis applied to CoP, posture, and muscle-activation data.
Design:
Participants included PD patients (n=72), healthy young adults (n=64), and healthy older adults (n=49). PD participants were stratified into mild (Hoehn & Yahr < 3) and severe (≥3) subgroups. During 30 s of quiet standing, CoP trajectories, full-body posture (depth sensor), and surface electromyography of TA and soleus were recorded. Forty variables quantifying CoP spatiotemporal and frequency features, joint angles and angular velocities, and muscle activity were extracted. EFA was performed, and factor scores were submitted to Gaussian mixture clustering.
Results: Nine factors were extracted, representing sway magnitude; mediolateral and anteroposterior frequency components; high-frequency sway with TA coupling; closed-loop control features; sway-velocity variability; sagittal- and frontal-plane posture; and joint angular velocity. Clustering identified seven distinct postural phenotypes. A cluster with globally low factor scores predominantly included healthy young adults, whereas a cluster with minor postural abnormalities but increased sway magnitude and TA activity included more healthy older adults. In contrast, clusters characterized by reduced sway but marked stooped posture, TA-dominant high-frequency activity, or severe multidimensional impairment were dominated by PD participants.
Conclusion:
Integrating CoP, posture, and muscle activation revealed PD-specific postural control mechanisms that cannot be explained by aging alone. This multidimensional, data-driven phenotyping approach supports individualized assessment and targeted rehabilitation strategies in PD.
