Introduction

Osteoarthritis (OA) of the knee is a common disease that impacts functional mobility and quality of life for many individuals [1]. Total Knee Arthroplasty (TKA) is the most common surgical treatment for end-stage OA. By replacing the impaired knee joint with implants, knee joint function and quality of life can be improved [2]. Since a primary goal of TKA is to regain ambulatory function, assessment of recovery after unilateral TKA using biomechanical gait variables can be useful clinical indicators. Monitoring the improvement of functional performance after TKA usually includes examining flexion and adduction knee moments and directional ground reaction forces [3]. Typically, studies on gait patterns before and after TKA use overground force plates and optical motion capture systems to collect biomechanical data [4]. While these instruments produce valid biomechanical data, they are constrained to laboratory settings and may not reflect real-world mobility patterns.
Wearable sensor-based continuous gait analysis is a promising alternative that can address the limitations of laboratory-based biomechanical evaluations [5]. Wearable devices are getting smaller and smarter, and wearables are easy to use and do not interfere with the natural behavior of the subject. Thus, wearable sensor-based gait analysis is one of the most promising methods used to quantify gait patterns in real-world conditions. Acceleration data from wearable sensors can also provide kinetic attributes outside of the laboratory setting. Monitoring multiple and sequential kinetic parameters using wearables in real-world conditions may lead to a more accurate understanding of the relationship between movement patterns and recovery from TKA.
Given the potential of wearable sensor-based gait analysis and the importance of gait outcomes post-TKA, the purpose of this study was to estimate biomechanical gait metrics using two ankle-worn wearable sensors in individuals after unilateral TKA. Biomechanical variables related to initial loading behavior were selected since they are correlated with OA progression [6]. Concurrently, multiple inertial gait variables including temporal and kinetic parameters were extracted using wearable sensors. Then, the inertial gait variables were used to develop statistical prediction models of the selected biomechanical gait variables, such as moments and ground reaction forces, in the TKA pĀ­opulation.
Two main contributions of the present work are: (1) a generic method for a linear inertial gait variable extraction, and (2) statistical models for estimation of important biomechanical gait variables.