Estimating Lower Extremity Running Gait Kinematics with a Single Accelerometer: A Deep Learning Approach

Abstract

Abnormal running kinematics are associated with an increased incidence of lower extremity injuries among runners. Accurate and unobtrusive running kinematic measurement plays an important role in the detection of gait abnormalities and the prevention of injuries among runners. Inertial-based methods have been proposed to address this need. However, previous methods require cumbersome sensor setup or participant-specific calibration. This study aims to validate a shoe-mounted accelerometer for sagittal plane lower extremity angle measurement during running based on a deep learning approach. A convolutional neural network (CNN) architecture was selected as the regression model to generalize in inter-participant scenarios and to minimize poorly estimated joints. Motion and accelerometer data were recorded from ten participants while running on a treadmill at five different speeds. The reference joint angles were measured by an optical motion capture system. The CNN model predictions deviated from the reference angles with a root mean squared error (RMSE) of less than 3.5° and 6.5° in intra- and inter-participant scenarios, respectively. Moreover, we provide an estimation of six important gait events with a mean absolute error of less than 2.5° and 6.5° in intra- and inter-participants scenarios, respectively. This study highlights an appealing minimal sensor setup approach for gait analysis purposes.

Combining Inertial Sensors and Machine Learning to Predict vGRF and Knee Biomechanics during a Double Limb Jump Landing Task

“…The only double limb task used was a two-leg jump, and reported percent difference between the expected and predicted peak vertical knee force was 22.9%, highlighting the challenges of estimating double limb tasks. Gholami et al [ 29 ] used a single accelerometer on the foot combined with a convolutional neural network (CNN) to predict lower extremity kinematics during running gait, but did not predict kinetics. In relation to the work cited above, our project strategy specifically targeted the task and biomechanical variables of interest in post-ACLR.…”

“…Additionally, the use of inertial sensors combined with machine learning to predict biomechanics in musculoskeletal-injured populations has grown exponentially in recent years [ 19 ], serving as a promising interdisciplinary solution. This combined approach has proven successful for predicting knee-specific biomechanics for single-limb tasks including running, jump landing, and cutting [ 27 , 28 , 29 ]. However, these solutions have not specifically targeted double limb landing tasks, which have clinical relevance post-ACLR.…”

Comparison between Accelerometer and Gyroscope in Predicting Level-Ground Running Kinematics by Treadmill Running Kinematics Using a Single Wearable Sensor

“…{r i | r i = p i −p i } where {p i } is the set of data points at specific gait events of the target signal (i.e., either at peak knee flexion during stance or peak hip flexion) andp i is the predicted value of p i in the regression. Following Gholami et al [5], in the calculation of RMSE, NRMSE, ST shown in Table 2, the peaks' positions indicated in Figure 2 were used, i.e., for the knee flexion and the maximum position in the hip flexion. For points in the curves were used.…”

Treadmill-to-Overground Mapping of Marker Trajectory for Treadmill-Based Continuous Gait Analysis

“…MoCap is often combined with a treadmill to achieve a continuous walking trajectory for biomechanical research in a limited space [ 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 ]. A treadmill provides a convenient experiment environment in terms of unlimited gait trackability under volume constraint and speed controllability.…”

“…Therefore, most of the gait data obtained through treadmill walking are analyzed based on gait cycle percentage. That is, each gait period from one heel strike to the next heel strike of the same foot is the time normalized as one gait cycle and the average and standard deviation of a physical quantity with respect to the gait cycle are used for the gait analysis [ 5 , 6 , 7 , 13 , 14 , 15 , 16 , 17 , 21 , 22 , 23 , 24 ].…”