Development of a wearable system to estimate knee adduction moment of patients with knee osteoarthritis during gait using a single inertial measurement unit

The external knee adduction moment (KAM) during gait is an important mechanical factor for knee osteoarthritis (KOA). Three-dimensional motion capture (MoCap) has been used to measure KAM, but it has limitations including cost, measuring time, and space. The purpose of this study was to develop a simple and wearable system to estimate KAM. Thirty-nine female and seven male subjects diagnosed with KOA were included. Simultaneous measurements of 5 ​m walk using MoCap and inertial measurement unit (IMU) attached to the knee were performed, and a deep learning algorithm based on a gait phase detection algorithm and one-dimensional convolutional neural network model was used to estimate KAM. In addition, the reliability of the artificial intelligence system was also evaluated. The mean absolute percent error (MAPE) and mean absolute error (MAE) of estimated KAM peak of the training data and test datasets were 13.0% and 0.0428 Nm.s/(BW∗Ht) and 17.3% and 0.0505 Nm.s/(BW∗Ht), respectively, and those of estimated KAM impulse were 20.4% and 0.0214 Nm.s/(BW∗Ht) and 21.4% and 0.0212 Nm.s/(BW∗Ht), respectively. The Pearson correlation coefficients between estimated KAM (peak, impulse) and MoCap KAM (peak, impulse) were 0.870 and 0.975 in the training dataset and 0.591 and 0.690 in the test dataset. The intraclass correlation coefficients (ICCs) of estimated KAM impulse were 0.83, 0.94, 0.70, and 0.82 for ICCs (1.1), (1,k), (2,1), and (2,k) respectively. The wearable system showed reasonable accuracy and reliability in estimating KAM of KOA during gait. The simple equipment in the system enables the measurement of KAM during daily practice in the clinic or hospital if there is available space to walk 5 ​m. • A wearable system estimates knee adduction moment (KAM) using a single sensor. • The system provides a cost-effective alternative to traditional motion capture. • Deep learning algorithms enhance the accuracy and reliability of KAM estimation. • The system allows KAM assessment in clinical settings with minimal equipment.