Researchers have identified a distinct biomechanical gait pattern in patients with knee osteoarthritis (OA) and high body mass index (BMI) using unsupervised machine learning. Published this week in Arthritis Care & Research, the study analyzed data from 431 participants enrolled in the Intensive Diet and Exercise for Arthritis trial to determine how gait biomechanics relate to the severity of the disease. The findings were released on November 5, 2026, offering new insights into limb-level loading profiles.
Identifying Biomechanical Phenotypes
The study cohort featured participants with a mean age of 65.6 years, a mean BMI of 33.6 kg/m², and an average walking speed of 1.2 m/s. Investigators utilized vertical ground reaction force waveforms and principal component analysis to evaluate shared variations between specific gait patterns and clinical markers, including inflammatory biomarkers, pain levels, physical function, and radiographic severity.
The machine learning analysis revealed a sustained loading pattern defined by reduced ground reaction forces during the early and late stages of the stance phase, contrasted by increased forces during the midstance. This specific pattern was linked to several negative outcomes:
- Higher inflammatory biomarkers
- Greater reported pain
- Worse Kellgren-Lawrence grades
- Poorer patient-reported function
- Slower walking speeds
Implications for Clinical Strategy
Beyond the sustained loading pattern, the research team identified a secondary profile characterized by early-to-midstance loading. This pattern was associated with female sex, higher BMI, increased levels of inflammation, and heightened radiographic severity. According to lead author Ashley N. Buck, PhD, of the University of Wisconsin-Madison, and her colleagues, these latent biomechanical phenotypes are fundamental to informing future rehabilitation strategies. As the authors noted, The specific limb-level loading profiles associated with clinically relevant knee OA features in individuals with high BMI remain unclear.
The researchers suggested that these gait loading patterns may reflect broader clinical and biological characteristics of knee OA that exceed the information provided by traditional peak joint loading measurements. Future clinical trials are expected to investigate whether targeting these specific waveform-based patterns can lead to meaningful improvements in joint tissue health, symptoms, and functional outcomes. The study data is officially cataloged under DOI:10.1002/acr.80186.
