To expand the potential use of in-shoe motion sensors (IMSs) in daily healthcare or activity monitoring applications for healthy subjects, we propose a real-time temporal estimation method for gait parameters concerning bilateral lower limbs (GPBLLs) that uses a single IMS and is based on a gait event detection approach. To validate the established methods, data from 26 participants recorded by an IMS and a reference 3D motion analysis system were compared. The agreement between the proposed method and the reference system was evaluated by the intraclass correlation coefficient (ICC). The results showed that, by averaging over five continuous effective strides, all time parameters achieved precisions of no more than 30 ms and agreement at the “excellent” level, and the symmetry indexes of the stride time and stance phase time achieved precisions of 1.0% and 3.0%, respectively, and agreement at the “good” level. These results suggest our method is effective and shows promise for wide use in many daily healthcare or activity monitoring applications for healthy subjects.
Frailty poses a threat to the daily lives of healthy older adults, highlighting the urgent need for technologies that can monitor and prevent its progression. Our objective is to demonstrate a method for providing long-term daily frailty monitoring using an in-shoe motion sensor (IMS). We undertook two steps to achieve this goal. Firstly, we used our previously established SPM-LOSO-LASSO (SPM: statistical parametric mapping; LOSO: leave-one-subject-out; LASSO: least absolute shrinkage and selection operator) algorithm to construct a lightweight and interpretable hand grip strength (HGS) estimation model for an IMS. This algorithm automatically identified novel and significant gait predictors from foot motion data and selected optimal features to construct the model. We also tested the robustness and effectiveness of the model by recruiting other groups of subjects. Secondly, we designed an analog frailty risk score that combined the performance of the HGS and gait speed with the aid of the distribution of HGS and gait speed of the older Asian population. We then compared the effectiveness of our designed score with the clinical expert-rated score. We discovered new gait predictors for HGS estimation via IMSs and successfully constructed a model with an “excellent” intraclass correlation coefficient and high precision. Moreover, we tested the model on separately recruited subjects, which confirmed the robustness of our model for other older individuals. The designed frailty risk score also had a large effect size correlation with clinical expert-rated scores. In conclusion, IMS technology shows promise for long-term daily frailty monitoring, which can help prevent or manage frailty for older adults.
Our purpose was to demonstrate the possibility of providing foot-healthcare application by using an in-shoe motion sensor (IMS) through validating the feasibility of applying an IMS for measuring the first metatarsophalangeal angle (FMTPA), which is the most important parameter regarding the common foot problem hallux valgus. Methods: The IMS signals can represent foot motions when the mid-foot and hindfoot were modelled as a rigid body. FMTPAs can be estimated from the foot-motion signals measured using an IMS embedded beneath the foot arch near the calcaneus side using a machine-learning method. The foot-motion signals were collected from 50 participants with different FMTPAs. The true FMTPAs were assessed from digital photography. Correlation-based feature-selection processes (significance level p < 0.05) were used to search for the predictors from the foot-motion signals. Leave-one-subject-out cross-validation, root mean squared error, and intra-class coefficients were used for FMTPA-estimation model evaluation. Results: Eleven FMTPA-impacted gait-phase clusters, which were used to construct effective foot-motion predictors, were observed in all gait-cycle periods except terminal swing. The range of the foot motion in the sagittal and coronal planes significantly correlated with the FMTPA (p < 0.05). Linear regression could be the best method for constructing an FMTPA estimation model with a root mean squared error and intra-class correlation coefficient of 4.2 degrees and 0.789, respectively. Conclusion: The results indicate the reliability of our FMTPA estimation model constructed from foot-motion signals and the possibility to providing foot-healthcare applications by using an IMS.
Mobility is a crucial physical factor that is highly correlated with fall risk and can be modified by physical exercise. The timed up and go (TUG) test outcome is a metric that represents mobility. Monitoring mobility in daily living is essential for fall prevention in older adults. In this study, we aimed to develop models for estimating TUG-represented mobility by a single stride of foot motion measured by an in-shoe motion sensor (IMS). Additionally, we tested the validity and individual robustness of these models on separately recruited subjects. As a result, we discovered novel gait predictors for high-precision estimation from single stride of foot motion. The models achieved an excellent intra-class correlation coefficient agreement between reference and estimated values with mean absolute errors (MAE) of 0.39 and 0.41 s (0.34 and 0.40 times of the standard deviation) for both male and female subjects. Around 80% of female subjects were effectively estimated within the limit of agreement of Bland-Altman plots, which suggested the individual robustness of the model.
Knee joint function deterioration significantly impacts quality of life. This study developed estimation models for ten knee indicators using data from in-shoe motion sensors to assess knee movement during everyday activities. Sixty-six healthy young participants were involved, and multivariate linear regression was employed to construct the models. The results showed that eight out of ten models achieved a “fair” to “good” agreement based on intra-class correlation coefficients (ICCs), with three knee joint angle indicators reaching the “fair” agreement. One temporal indicator model displayed a “good” agreement, while another had a “fair” agreement. For the angular jerk cost indicators, three out of four attained a “fair” or “good” agreement. The model accuracy was generally acceptable, with the mean absolute error ranging from 0.54 to 0.75 times the standard deviation of the true values and errors less than 1% from the true mean values. The significant predictors included the sole-to-ground angles, particularly the foot posture angles in the sagittal and frontal planes. These findings support the feasibility of estimating knee function solely from foot motion data, offering potential for daily life monitoring and rehabilitation applications. However, discrepancies in the two models were influenced by the variance in the baseline knee flexion and sensor placement. Future work will test these models on older and osteoarthritis-affected individuals to evaluate their broader applicability, with prospects for user-tailored rehabilitation applications. This study is a step towards simplified, accessible knee health monitoring through wearable technology.
Purpose Cervical myelopathy (CM) often presents with gait disturbances, which can lead to falls and spinal cord injury. As the presence of gait disturbances negatively affects postoperative improvement in patients with CM, early detection of gait disturbances is crucial. However, current gait analysis methods require specialisation, limiting their feasibility for screening. This study aimed to perform gait analysis in patients with CM using wearable in-shoe inertial measurement unit (IMU) sensors and to develop a screening system using machine learning. Methods This study included 26 patients with CM and 26 age-matched healthy controls. Gait parameters were measured using IMU sensors while participants walked along a 16 m corridor. Eleven gait parameters and their coefficients of variation (CV) were analysed. A machine learning model using the extreme gradient boosting algorithm was developed for CM classification. Results Patients with CM exhibited significantly lower gait speed, stride length, dorsiflexion angle, plantarflexion angle, and foot height, along with increased stance time and stance phase. CVs of gait parameters were higher in patients with CM, with an increased stance phase observed from the early stages. The classification model achieved high performance (AUC = 0.89, sensitivity = 81%, specificity = 85%), with stance phase and CVs of stride length and plantarflexion angle identified as key predictive features. Conclusion Wearable IMU sensors effectively captured gait disturbances in CM patients, and machine learning enabled accurate classification. This method enables daily gait monitoring and could facilitate early detection and intervention for CM.