The traditional evaluation of gait in the laboratory during structured testing has provided important insights, but is limited by its “snapshot” character and observation in an unnatural environment. Wearables enable monitoring of gait in real-world environments over a week. Initial findings show that in-lab and real-world measures differ. As a step towards better understanding these gaps, we directly compared in-lab usual-walking (UW) and dual-task walking (DTW) to daily-living measures of gait. In-lab gait features (e.g., gait speed, step regularity, and stride regularity) derived from UW and DTW were compared to the same gait features during daily-living in 150 elderly fallers (age: 76.5 ± 6.3 years, 37.6% men). In both settings, features were extracted from a lower-back accelerometer. In the real-world setting, subjects were asked to wear the device for 1 week and pre-processing detected 30-s daily-living walking bouts. A histogram of all walking bouts was determined for each walking feature for each subject and then each subject’s typical (percentile 50, median), worst (percentile 10) and the best (percentile 90) values over the week were determined for each feature. Statistics of reliability were assessed using Intra-Class correlations and Bland-Altman plots. As expected, in-lab gait speed, step regularity, and stride regularity were worse during DTW, compared to UW. In-lab gait speed, step regularity, and stride regularity during UW were significantly higher (i.e., better) than the typical daily-living values (p < 0.001) and different (p < 0.001) from the worst and best values. DTW values tended to be similar to typical daily-living values (p = 0.205, p = 0.053, p = 0.013 respectively). ICC assessment and Bland-Altman plots indicated that in-lab values do not reliably reflect the daily-walking values. Gait values measured during relatively long (30-s) daily-living walking bouts are more similar to the corresponding values obtained in the lab during dual-task walking, as compared to usual walking. Still, gait performance during most daily-living walking bouts is worse than that measured during usual and dual-tasking in the lab. The values measured in the lab do not reliably reflect daily-living measures. That is, an older adult’s typical daily-living gait cannot be estimated by simply measuring walking in a structured, laboratory setting.

The step-by-step determination of the spatio-temporal parameters of gait is clinically relevant since it provides an estimation of the variability of specific gait patterns associated with frequent geriatric syndromes. In recent years, several methods, based on the use of magneto-inertial units (MIMUs), have been developed for the step-by-step estimation of the gait temporal parameters. However, most of them were applied to the gait of healthy subjects and/or of a single pathologic population. Moreover, spatial parameters in pathologic populations have been rarely estimated step-by-step using MIMUs. The validity of clinically suitable MIMU-based methods for the estimation of spatio-temporal parameters is therefore still an open issue. The aim of this study was to propose and validate a method for the determination of both temporal and spatial parameters that could be applied to normal and heavily compromised gait patterns. Methods: Two MIMUs were attached above each subject's ankles. An instrumented gait mat was used as gold standard. Gait data were acquired from ten hemiparetic subjects, ten choreic subjects, ten subjects with Parkinson's disease and ten healthy older adults walking at two different gait speeds. The method detects gait events (GEs) taking advantage of the cyclic nature of gait and exploiting some lower limb invariant kinematic characteristics. A combination of a MIMU axes realignment along the direction of progression and of an optimally filtered direct and reverse integration is used to determine the stride length. Results: Over the 4,514 gait cycles analyzed, neither missed nor extra GEs were generated. The errors in identifying both initial and final contact at comfortable speed ranged between 0 and 11 ms for the different groups analyzed. The stride length was estimated for all subjects with less than 3% error. Conclusions: The proposed method is apparently extremely robust since gait speed did not substantially affect its performance and both missed and extra GEs were avoided. The spatio-temporal parameters estimates showed smaller errors than those reported in previous studies and a similar level of precision and accuracy for both healthy and pathologic gait patterns. The combination of robustness, precision and accuracy suggests that the proposed method is suitable for routine clinical use.

Machine learning methods have been widely used for gait assessment through the estimation of spatio-temporal parameters. As a further step, the objective of this work is to propose and validate a general probabilistic modeling approach for the classification of different pathological gaits. Specifically, the presented methodology was tested on gait data recorded on two pathological populations (Huntington’s disease and post-stroke subjects) and healthy elderly controls using data from inertial measurement units placed at shank and waist. By extracting features from group-specific Hidden Markov Models (HMMs) and signal information in time and frequency domain, a Support Vector Machines classifier (SVM) was designed and validated. The 90.5% of subjects was assigned to the right group after leave-one-subject–out cross validation and majority voting. The long-term goal we point to is the gait assessment in everyday life to early detect gait alterations.

Background: The estimation of the spatio-temporal gait parameters is of primary importance in both physical activity monitoring and clinical contexts. A method for estimating step length bilaterally, during level walking, using a single inertial measurement unit (IMU) attached to the pelvis is proposed. In contrast to previous studies, based either on a simplified representation of the human gait mechanics or on a general linear regressive model, the proposed method estimates the step length directly from the integration of the acceleration along the direction of progression. Methods: The IMU was placed at pelvis level fixed to the subject's belt on the right side. The method was validated using measurements from a stereo-photogrammetric system as a gold standard on nine subjects walking ten laps along a closed loop track of about 25 m, varying their speed. For each loop, only the IMU data recorded in a 4 m long portion of the track included in the calibrated volume of the SP system, were used for the analysis. The method takes advantage of the cyclic nature of gait and it requires an accurate determination of the foot contact instances. A combination of a Kalman filter and of an optimally filtered direct and reverse integration applied to the IMU signals formed a single novel method (Kalman and Optimally filtered Step length Estimation - KOSE method). A correction of the IMU displacement due to the pelvic rotation occurring in gait was implemented to estimate the step length and the traversed distance. Results: The step length was estimated for all subjects with less than 3% error. Traversed distance was assessed with less than 2% error. Conclusions: The proposed method provided estimates of step length and traversed distance more accurate than any other method applied to measurements obtained from a single IMU that can be found in the literature. In healthy subjects, it is reasonable to expect that, errors in traversed distance estimation during daily monitoring activity would be of the same order of magnitude of those presented.

Quantitative gait analysis can provide a description of joint kinematics and dynamics, and it is recognized as a clinically useful tool for functional assessment, diagnosis and intervention planning. Clinically interpretable parameters are estimated from quantitative measures (i.e. ground reaction forces, skin marker trajectories, etc.) through biomechanical modelling. In particular, the estimation of joint moments during motion is grounded on several modelling assumptions: (1) body segmental and joint kinematics is derived from the trajectories of markers and by modelling the human body as a kinematic chain; (2) joint resultant (net) loads are, usually, derived from force plate measurements through a model of segmental dynamics. Therefore, both measurement errors and modelling assumptions can affect the results, to an extent that also depends on the characteristics of the motor task analysed (i.e. gait speed). Errors affecting the trajectories of joint centres, the orientation of joint functional axes, the joint angular velocities, the accuracy of inertial parameters and force measurements (concurring to the definition of the dynamic model), can weigh differently in the estimation of clinically interpretable joint moments. Numerous studies addressed all these methodological aspects separately, but a critical analysis of how these aspects may affect the clinical interpretation of joint dynamics is still missing. This article aims at filling this gap through a systematic review of the literature, conducted on Web of Science, Scopus and PubMed. The final objective is hence to provide clear take-home messages to guide laboratories in the estimation of joint moments for the clinical practice.