Joint kinematics alone can distinguish hip or knee osteoarthritis patients from asymptomatic controls with high accuracy

Abstract

Osteoarthritis (OA) is one of the leading musculoskeletal disabilities worldwide, and several interventions intend to change the gait pattern in OA patients to more healthy patterns. However, an accessible way to follow up the biomechanical changes in a clinical setting is still missing. Therefore, this study aims to evaluate whether we can use biomechanical data collected from a specific activity of daily living to help distinguish hip OA patients from controls and knee OA patients from controls using features that potentially could be measured in a clinical setting. To achieve this goal, we considered three different classes of statistical models with different levels of data complexity. Class 1 is kinematics based only (clinically applicable), class 2 includes joint kinetics (semi‐applicable under the condition of access to a force plate or prediction models), and class 3 uses data from advanced musculoskeletal modeling (not clinically applicable). We used a machine learning pipeline to determine which classification model was best. We found 100% classification accuracy for KneeOA‐vs‐Asymptomatic and 93.9% for HipOA‐vs‐Asymptomatic using seven features derived from the lumbar spine and hip kinematics collected during ascending stairs. These results indicate that kinematical data alone can distinguish hip or knee OA patients from asymptomatic controls. However, to enable clinical use, we need to validate if the classifier also works with sensor‐based kinematical data and whether the probabilistic outcome of the logistic regression model can be used in the follow‐up of patients with OA.

Can the Output of a Learned Classification Model Monitor a Person’s Functional Recovery Status Post-Total Knee Arthroplasty?

“…Based on Emmerzaal et al, (2022) ascending stairs, descending stairs, and level walking are the three activities that best distinguish people with end-stage knee OA from asymptomatic controls [ 13 ]. Therefore, the participants were asked to perform these three activities (i.e., level walking, ascending, and descending stairs), while being equipped with 17 inertial measurement units (IMUs) (MVN BIOMECH Awinda Xsens Technologies, sampling at 60 Hz; software version MVN Awinda 4.4) [ 17 ].…”

“…Joint angles were directly extracted from the Xsens graphical user interface (MVN Awinda studio 4.4) [ 17 ]. Similar to the workflow of Emmerzaal et al (2022), only a single middle stride was extracted for each leg and used for further processing [ 13 ]. Step detection for walking was performed using the peak acceleration of the anterior-posterior raw acceleration signal [ 19 ].…”

“…The peak tibia acceleration was found to be the most robust method to segment stairs data based on the comparison of the step detection based on the peaks in the acceleration signal with the step detection using force plate data. Joint angles from the spine, hip, knee, and ankle of the middle stride were extracted and filtered at 6 Hz to match conventional filtering of joint kinematics used in our previous work [ 13 ] and time normalised to 100% stride time.…”

“…Previous studies (Laroche et al, 2014 or Jones et al, 2006) used machine learning in osteoarthritis research to distinguish an osteoarthritic biomechanical gait pattern from the gait pattern of an asymptomatic control subject [ 11 , 12 ]. Recently, we used an L1-regularised logistic regression model to distinguish people with (either pre-operative hip or knee) OA from asymptomatic controls [ 13 ]. Separate models were created for five activities of daily living and four rehabilitation exercises.…”

“…Separate models were created for five activities of daily living and four rehabilitation exercises. Marker-based joint kinematics during ascending stairs, descending stairs, and level walking could accurately classify people with knee OA and asymptomatic controls [ 13 ].…”