2025 ISAKOS Biennial Congress In-Person Poster
Movement Profiles in Patellofemoral Pain Patients During the Lateral Step-Down Task: an Unsupervised Machine Learning Approach
Enzo Salviato Mameri, MD, MSc, São Paulo, São Paulo BRAZIL
Leonardo Metsavaht, MD PhD, Rio de Janeiro, RJ BRAZIL
Eliane C Guadagnin, PhD, São Paulo BRAZIL
Felipe Gonzalez, MD, Chicago UNITED STATES
Jonathan A Gustafson, PhD, Chicago, Illinois UNITED STATES
Thais Dutra Vieira, MD, Lyon, Rhone FRANCE
Jorge Chahla, MD, PhD, Hinsdale, IL UNITED STATES
Bertrand Sonnery-Cottet, MD, PhD, Lyon, Rhône FRANCE
Marcus V. M. Luzo, MD, PhD, São Paulo, SP BRAZIL
Gustavo Leporace, PhD, Rio De Janeiro, Rio de Janeiro BRAZIL
Instituto Brasil de Tecnologia da Saúde (IBTS), Rio de Janeiro, Rio de Janeiro, BRAZIL
FDA Status Not Applicable
Summary
Patients with patellofemoral pain exhibit four distinct clinically relevant subgroups based on their kinematics during a step-down task, with a protective movement profile associated with worse pain and function.
Abstract
Background
In the pursuit of more effective treatment approaches for patellofemoral pain, targeted interventions for specific subgroups of individuals have been proposed in the literature. Nevertheless, there is a lack of consensus regarding the optimal identification of these subgroups to guide effective management strategies.
Objective
The aim of this study was to investigate the existence of different motion profiles among individuals with patellofemoral pain during the lateral step-down task and compare their clinical and physical characteristics.
Methods
Individuals with patellofemoral pain had their three-dimensional kinematics assessed during a lateral step-down task using an optoelectronic system. The variables analyzed were peak trunk flexion, lateral trunk tilt, anterior pelvic tilt, contralateral pelvic drop, hip adduction, hip internal rotation, and knee flexion. Self-Organizing Maps (SOM) and K-means clustering techniques were used to identify distinct biomechanical profiles. Clinical characteristics compared among profiles were hip and knee strength (using hand-held dynamometers) and range of motion (digital inclinometer), as well as patient-reported outcome measures (IKDC, LEFS, VAS pain, Tampa Scale for Kinesiophobia). Patient-reported scores, strength metrics, range of motion, age, body mass, height, and body mass index (BMI) were treated as continuous variables and compared among groups using the Kruskal-Wallis test. The Tukey test was also used for multiple group comparisons. Differences in gender distribution among groups were assessed using the Chi-Square test. The significance level was set at 5%.
Results
The average age of the 49 participants was 41.6 ± 13.5 years, with a BMI of 24.3 ± 4.3 kg/m2. Among them, 61.2% were female (n=30). Four motion profiles were identified for the lateral step-down task in individuals with patellofemoral pain. Profile 1 (Balanced Alignment Profile) exhibited a trunk-hip-knee aligned movement pattern, with increased knee flexion. This profile presented greater IKDC (p=0.01) and smaller VAS-pain scores (p=0.02). Profiles 2 (Trunk-hip-knee Imbalance Profile) and 3 (Pelvic-Hip Interactor Profile) depicted increased dynamic knee valgus. However, profile 2 presented limited trunk and knee flexion, while profile 3 presented increased anterior pelvic tilt and trunk flexion. Profile 2 had excessive passive hip internal rotation and was predominantly female, while profile 3 exhibited increased hip and knee strength and lower levels of pain. Profile 4 (Protective Movement Profile) exhibited a protective adaptation, showing decreased peaks for all variables and increased lateral trunk tilt, as well as the lowest IKDC scores (p=0.01) and highest VAS-pain scores (p=0.02). No significant differences were detected for LEFS (p=0.28) or the Tampa Scale of Kinesiophobia (p=0.62). Regarding gender distribution, statistically significant differences were observed (p=0.003). Profile 1 contained fewer women than expected, whereas profile 2 had fewer men. There were no significant differences in age (p=0.45), body mass (p=0.166), height (p=0.07), or BMI (p=0.321) across profiles.
Conclusion
There are four clinically relevant subgroups of individuals with patellofemoral pain based on their kinematics during a step-down task, with a protective movement profile associated with worse pain and function. The associated modifiable characteristics identified here can provide valuable insights for clinicians to implement targeted interventions and improve patient care.