Pre-operative sensor-based gait parameters predict functional outcome after total knee arthroplasty

Background: Despite the general success of total knee arthroplasty (TKA) regarding patient-reported outcome measures, studies investigating gait function have shown diverse functional outcomes. Mobile sensor-based systems have recently been employed for accurate clinical gait assessments, as they allow a better integration of gait analysis into clinical routines as compared to laboratory based systems. Research question: In this study, we sought to examine whether an accurate assessment of gait function of knee osteoarthritis patients with respect to surgery outcome evaluation after TKA using a mobile sensor-based gait analysis system is possible. Methods: A foot-worn sensor-based system was used to assess spatio-temporal gait parameters of 24 knee osteoarthritis patients one day before and one year after TKA, and in comparison to matched control participants. Patients were clustered into positive and negative responder groups using a heuristic approach regarding improvements in gait function. Machine learning was used to predict surgery outcome based on pre-operative gait parameters. Results: Gait function differed significantly between controls and patients. Patient-reported outcome measures improved significantly after surgery, but no significant global gait parameter difference was observed between pre- and post-operative status. However, the responder groups could be correctly predicted with an accuracy of up to 89% using pre-operative gait parameters. Patients exhibiting high pre-operative gait function were more likely to experience a functional decrease after surgery. Important gait parameters for the discrimination were stride time and stride length. Significance: The early identification of post-surgical functional outcomes of patients is of great importance to better inform patients pre-operatively regarding surgery success and to improve post-surgical management.