What is the Best Configuration of Wearable Sensors to Measure Spatiotemporal Gait Parameters in Children with Cerebral Palsy?

Wearable Technology and Its Influence on Motor Development and Biomechanical Analysis

The convergence among biomechanics, motor development, and wearable technology redefines our understanding of human movement. These technologies allow for the continuous monitoring of motor development and the state of motor abilities from infancy to old age, enabling early and personalized interventions to promote healthy motor skills. For athletes, they offer valuable insights to optimize technique and prevent injuries, while in old age, they help maintain mobility and prevent falls. Integration with artificial intelligence further extends these capabilities, enabling sophisticated data analysis. Wearable technology is transforming the way we approach motor development and maintenance of motor skills, offering unprecedented possibilities for improving health, performance, and quality of life at every stage of life. The promising future of these technologies paves the way for an era of more personalized and effective healthcare, driven by innovation and interdisciplinary collaboration.

Validation of an algorithm to assess regular and irregular gait using inertial sensors in healthy and stroke individuals

Background

Studies using inertial measurement units (IMUs) for gait assessment have shown promising results regarding accuracy of gait event detection and spatiotemporal parameters. However, performance of such algorithms is challenged in irregular walking patterns, such as in individuals with gait deficits. Based on the literature, we developed an algorithm to detect initial contact (IC) and terminal contact (TC) and calculate spatiotemporal gait parameters. We evaluated the validity of this algorithm for regular and irregular gait patterns against a 3D optical motion capture system (OMCS).

Methods

Twenty healthy participants (aged 59 ± 12 years) and 10 people in the chronic phase after stroke (aged 61 ± 11 years) were equipped with 4 IMUs: on both feet, sternum and lower back (MTw Awinda, Xsens) and 26 reflective makers. Participants walked on an instrumented treadmill for 2 minutes (i) with their preferred stride lengths and (ii) once with irregular stride lengths (±20% deviation) induced by light projected stepping stones. Accuracy of the algorithm was evaluated on stride-by-stride agreement of IC, TC, stride time, length and velocity with OMCS. Bland-Altman-like plots were made for the spatiotemporal parameters, while differences in detection of IC and TC time instances were shown in histogram plots. Performance of the algorithm was compared between regular and irregular gait with a linear mixed model. This was done by comparing the performance in healthy participants in the regular vs irregular walking condition, and by comparing the agreement in healthy participants with stroke participants in the regular walking condition.

Results

For each condition at least 1,500 strides were included for analysis. Compared to OMCS, IMU-based IC detection in both groups and condition was on average 9-17 (SD ranging from 7 to 35) ms, while IMU-based TC was on average 15-24 (SD ranging from 12 to 35) ms earlier. When comparing regular and irregular gait in healthy participants, the difference between methods was 2.5 ms higher for IC, 3.4 ms lower for TC, 0.3 cm lower for stride length, and 0.4 cm/s higher for stride velocity in the irregular walking condition. No difference was found on stride time. When comparing the differences between methods between healthy and stroke participants, the difference between methods was 7.6 ms lower for IC, 3.8 cm lower for stride length, and 3.4 cm/s lower for stride velocity in stroke participants. No differences were found on differences between methods on TC detection and stride time between stroke and healthy participants.

Conclusions

Small irrelevant differences were found on gait event detection and spatiotemporal parameters due to irregular walking by imposing irregular stride lengths or pathological (stroke) gait. Furthermore, IMUs seem equally good compared to OMCS to assess gait variability based on stride time, but less accurate based on stride length.

Walking Bout Detection for People Living in Long Residential Care: A Computationally Efficient Algorithm for a 3-Axis Accelerometer on the Lower Back

Accurate and reliable measurement of real-world walking activity is clinically relevant, particularly for people with mobility difficulties. Insights on walking can help understand mobility function, disease progression, and fall risks. People living in long-term residential care environments have heterogeneous and often pathological walking patterns, making it difficult for conventional algorithms paired with wearable sensors to detect their walking activity. We designed two walking bout detection algorithms for people living in long-term residential care. Both algorithms used thresholds on the magnitude of acceleration from a 3-axis accelerometer on the lower back to classify data as "walking" or "non-walking". One algorithm had generic thresholds, whereas the other used personalized thresholds. To validate and evaluate the algorithms, we compared the classifications of walking/non-walking from our algorithms to the real-time research assistant annotated labels and the classification output from an algorithm validated on a healthy population. Both the generic and personalized algorithms had acceptable accuracy (0.83 and 0.82, respectively). The personalized algorithm showed the highest specificity (0.84) of all tested algorithms, meaning it was the best suited to determine input data for gait characteristic extraction. The developed algorithms were almost 60% quicker than the previously developed algorithms, suggesting they are adaptable for real-time processing.

Enhancing Wearable Gait Monitoring Systems: Identifying Optimal Kinematic Inputs in Typical Adolescents

Machine learning-based gait systems facilitate the real-time control of gait assistive technologies in neurological conditions. Improving such systems needs the identification of kinematic signals from inertial measurement unit wearables (IMUs) that are robust across different walking conditions without extensive data processing. We quantify changes in two kinematic signals, acceleration and angular velocity, from IMUs worn on the frontal plane of bilateral shanks and thighs in 30 adolescents (8-18 years) on a treadmills and outdoor overground walking at three different speeds (self-selected, slow, and fast). Primary curve-based analyses included similarity analyses such as cosine, Euclidean distance, Poincare analysis, and a newly defined bilateral symmetry dissimilarity test (BSDT). Analysis indicated that superior-inferior shank acceleration (SI shank Acc) and medial-lateral shank angular velocity (ML shank AV) demonstrated no differences to the control signal in BSDT, indicating the least variability across the different walking conditions. Both SI shank Acc and ML shank AV were also robust in Poincare analysis. Secondary parameter-based similarity analyses with conventional spatiotemporal gait parameters were also performed. This normative dataset of walking reports raw signal kinematics that demonstrate the least to most variability in switching between treadmill and outdoor walking to help guide future machine learning models to assist gait in pediatric neurological conditions.

Relationship between spatiotemporal parameters and clinical outcomes in children with bilateral spastic cerebral palsy: Clinical interpretation proposal

BACKGROUND

Understanding the links between gait disorders, impairments, and activity limitations is essential for correctly interpreting the instrumented gait analysis. We aimed to evaluate the relationships between spatiotemporal parameters and clinical outcomes in children with bilateral spastic cerebral palsy, and find out whether spatiotemporal parameters provide clinical information regarding gait pattern and walking.

METHODS

Data from 19 children with bilateral spastic cerebral palsy (nine males, ten females, 9.6 ± 2.8 years old) were collected retrospectively. All children underwent an instrumented gait analysis and a standardized clinical assessment. Seven spatiotemporal parameters were calculated: non-dimensional cadence, stride length, step width, gait speed, first double support, single support, and time of toe off. Clinical outcomes included measures of two different components of the International Classification of Functioning, Disability and Health - Children and Youth version: body functions and structures (spasticity, contractures and range of motion, and deformities), and activities and participation (gross motor function, and walking capacity). Pearson correlation, ANOVA, Student's t, Mann-Whitney U, and Kruskal-Wallis tests were used to analyze relationships. Spatiotemporal parameters related to clinical outcomes of body functions and structures were interpreted as outcome measures of gait pattern, while those related to clinical outcomes of activities and participation were interpreted as outcome measures of walking.

RESULTS

Non-dimensional cadence, stride length, and gait speed showed relationships (p < 0.05) with hip flexors spasticity and hindfoot deformity, ankle plantar flexors spasticity, and hindfoot deformity, respectively. All spatiotemporal parameters except non-dimensional cadence showed correlation (p < 0.05) with gross motor function and walking capacity.

CONCLUSIONS

Spatiotemporal parameters provide clinical information regarding both gait pattern and walking.

Inertial sensors for gait monitoring and design of adaptive controllers for exoskeletons after stroke: a feasibility study

Introduction: Tuning the control parameters is one of the main challenges in robotic gait therapy. Control strategies that vary the control parameters based on the user's performance are still scarce and do not exploit the potential of using spatiotemporal metrics. The goal of this study was to validate the feasibility of using shank-worn Inertial Measurement Units (IMUs) for clinical gait analysis after stroke and evaluate their preliminary applicability in designing an automatic and adaptive controller for a knee exoskeleton (ABLE-KS). Methods: First, we estimated the temporal (i.e., stride time, stance, and swing duration) and spatial (i.e., stride length, maximum vertical displacement, foot clearance, and circumduction) metrics in six post-stroke participants while walking on a treadmill and overground and compared these estimates with data from an optical motion tracking system. Next, we analyzed the relationships between the IMU-estimated metrics and an exoskeleton control parameter related to the peak knee flexion torque. Finally, we trained two machine learning algorithms, i.e., linear regression and neural network, to model the relationship between the exoskeleton torque and maximum vertical displacement, which was the metric that showed the strongest correlations with the data from the optical system [r = 0.84; ICC(A,1) = 0.73; ICC(C,1) = 0.81] and peak knee flexion torque (r = 0.957). Results: Offline validation of both neural network and linear regression models showed good predictions (R2 = 0.70-0.80; MAE = 0.48-0.58 Nm) of the peak torque based on the maximum vertical displacement metric for the participants with better gait function, i.e., gait speed > 0.7 m/s. For the participants with worse gait function, both models failed to provide good predictions (R2 = 0.00-0.19; MAE = 1.15-1.29 Nm) of the peak torque despite having a moderate-to-strong correlation between the spatiotemporal metric and control parameter. Discussion: Our preliminary results indicate that the stride-by-stride estimations of shank-worn IMUs show potential to design automatic and adaptive exoskeleton control strategies for people with moderate impairments in gait function due to stroke.

Artificial Neural Network-Based Activities Classification, Gait Phase Estimation, and Prediction

Gait patterns are critical to health monitoring, gait impairment assessment, and wearable device control. Unrhythmic gait pattern detection under community-based conditions is a new frontier in this area. The present paper describes a high-accuracy gait phase estimation and prediction algorithm built on a two-stage artificial neural network. This work targets to develop an algorithm that can estimate and predict the gait cycle in real time using a portable controller with only two IMU sensors (one on each thigh) in the community setting. Our algorithm can detect the gait phase in unrhythmic conditions during walking, stair ascending, and stair descending, and classify these activities with standing. Moreover, our algorithm is able to predict both future intra- and inter-stride gait phases, offering a potential means to improve wearable device controller performance. The proposed data-driven algorithm is based on a dataset consisting of 5 able-bodied subjects and validated on 3 different able-bodied subjects. Under unrhythmic activity situations, validation shows that the algorithm can accurately identify multiple activities with 99.55% accuracy, and estimate ([Formula: see text]: 6.3%) and predict 200-ms-ahead ([Formula: see text]: 8.6%) the gait phase percentage in real time, which are on average 57.7 and 54.0% smaller than the error from the event-based method in the same conditions. This study showcases a solution to estimate and predict gait status for multiple unrhythmic activities, which may be deployed to controllers for wearable robots or health monitoring devices.

Accuracy and comparison of sensor-based gait speed estimations under standardized and daily life conditions in children undergoing rehabilitation

Background

Gait speed is a widely used outcome measure to assess the walking abilities of children undergoing rehabilitation. It is routinely determined during a walking test under standardized conditions, but it remains unclear whether these outcomes reflect the children's performance in daily life. An ankle-worn inertial sensor provides a usable opportunity to measure gait speed in the children's habitual environment. However, sensor-based gait speed estimations need to be accurate to allow for comparison of the children's gait speed between a test situation and daily life. Hence, the first aim of this study was to determine the measurement error of a novel algorithm that estimates gait speed based on data of a single ankle-worn inertial sensor in children undergoing rehabilitation. The second aim of this study was to compare the children’s gait speed between standardized and daily life conditions.

Methods

Twenty-four children with walking impairments completed four walking tests at different speeds (standardized condition) and were monitored for one hour during leisure or school time (daily life condition). We determined accuracy by comparing sensor-based gait speed estimations with a reference method in both conditions. Eventually, we compared individual gait speeds between the two conditions.

Results

The measurement error was 0.01 ± 0.07 m/s under the standardized and 0.04 ± 0.06 m/s under the daily life condition. Besides, the majority of children did not use the same speed during the test situation as in daily life.

Conclusion

This study demonstrates an accurate method to measure children's gait speed during standardized walking tests and in the children's habitual environment after rehabilitation. It only requires a single ankle sensor, which potentially increases wearing time and data quality of measurements in daily life. We recommend placing the sensor on the less affected side, unless the child wears one orthosis. In this latter case, the sensor should be placed on the side with the orthosis. Moreover, this study showed that most children did not use the same speed in the two conditions, which encourages the use of wearable inertial sensors to assess the children's walking performance in their habitual environment following rehabilitation.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12984-022-01079-3.

Feasibility of wearable technology for 'real-world' gait analysis in children with Prader-Willi and Angelman syndromes

BACKGROUND

Prader-Willi syndrome (PWS) and Angelman syndrome (AS) are neurodevelopmental disorders in need of innovative 'real-world' outcome measures to evaluate treatment effects. Instrumented gait analysis (IGA) using wearable technology offers a potentially feasible solution to measure "real-world' neurological and motor dysfunction in these groups.

METHODS

Children (50% female; 6-16 years) diagnosed with PWS (n = 9) and AS (n = 5) completed 'real-world' IGA assessments using the Physilog®5 wearable. PWS participants completed a laboratory assessment and a 'real-world' long walk. The AS group completed 'real-world' caregiver-assisted assessments. Mean and variability results for stride time, cadence, stance percentage (%) and stride length were extracted and compared across three different data reduction protocols.

RESULTS

The wearables approach was found to be feasible, with all participants able to complete at least one assessment. This study also demonstrated significant agreement, using Lin's concordance correlation coefficient (CCC), between laboratory and 'real-world' assessments in the PWS group for mean stride length, mean stance % and stance % CV (n = 7, CCC: 0.782-0.847, P = 0.011-0.009).

CONCLUSION

'Real-world' gait analysis using the Physilog®5 wearable was feasible to efficiently assess neurological and motor dysfunction in children affected with PWS and AS.

A Deep Learning Approach for Gait Event Detection from a Single Shank-Worn IMU: Validation in Healthy and Neurological Cohorts

Many algorithms use 3D accelerometer and/or gyroscope data from inertial measurement unit (IMU) sensors to detect gait events (i.e., initial and final foot contact). However, these algorithms often require knowledge about sensor orientation and use empirically derived thresholds. As alignment cannot always be controlled for in ambulatory assessments, methods are needed that require little knowledge on sensor location and orientation, e.g., a convolutional neural network-based deep learning model. Therefore, 157 participants from healthy and neurologically diseased cohorts walked 5 m distances at slow, preferred, and fast walking speed, while data were collected from IMUs on the left and right ankle and shank. Gait events were detected and stride parameters were extracted using a deep learning model and an optoelectronic motion capture (OMC) system for reference. The deep learning model consisted of convolutional layers using dilated convolutions, followed by two independent fully connected layers to predict whether a time step corresponded to the event of initial contact (IC) or final contact (FC), respectively. Results showed a high detection rate for both initial and final contacts across sensor locations (recall ≥92%, precision ≥97%). Time agreement was excellent as witnessed from the median time error (0.005 s) and corresponding inter-quartile range (0.020 s). The extracted stride-specific parameters were in good agreement with parameters derived from the OMC system (maximum mean difference 0.003 s and corresponding maximum limits of agreement (−0.049 s, 0.051 s) for a 95% confidence level). Thus, the deep learning approach was considered a valid approach for detecting gait events and extracting stride-specific parameters with little knowledge on exact IMU location and orientation in conditions with and without walking pathologies due to neurological diseases.

Validation of shoe-worn Gait Up Physilog®5 wearable inertial sensors in adolescents

BACKGROUND

Gait Up Physilog® wearable inertial sensors are a powerful alternative to traditional laboratory-based gait assessment for children with gait impairment. To build clinician trust in these devices and ultimately facilitate their use outside confined spaces, studies have examined performance of previous versions of Physilog® wearable inertial sensors but predominant focus has been on older adults. Despite their different gait patterns and behavioural/cognitive profiles, there are limited studies in children.

RESEARCH QUESTION

To determine whether key spatiotemporal gait parameters (stride length, time and velocity) collected by shoe-worn Physilog®5 sensors in a hallway assessment protocol are a valid method of gait assessment in typically developing adolescents aged 12-15 years.

METHODS

A total 30 typically developing participants (50 % female) median age 13.7 (interquartile range 2.34) were assessed in an exploratory study whilst walking at self-selected speed over the GAITRite® electronic walkway, concurrently wearing Physilog®5 sensors. Concurrent validity was analysed by Lin's concordance correlation coefficient (CCC), Bland-Altman plots and 95 % limit of agreement. Systematic bias was assessed using 95 % confidence interval of the mean difference.

RESULTS

Mean stride data demonstrated substantial agreement for stride length (CCC = 0.975) and stride velocity (CCC = 0.979) to almost perfect agreement for stride time (CCC > 0.996). Agreement between the technologies for individual stride-to-stride data remained high for stride time (CCC = 0.952); yet reduced for stride length (CCC = 0.868) and stride velocity (CCC = 0.877). Male/female differences in performance of the technology were observed for stride velocity, favouring females.

SIGNIFICANCE

Physilog®5 inertial sensors accurately measure walking in adolescents, with stride time the most accurately detected parameter. This demonstrates that wearables can be used by researchers and clinicians working with adolescent groups as an alternative to fixed systems. These findings will ultimately pave the way to using wearables for assessments with children outside of the laboratory environment.

Gait event detection using kinematic data in children with bilateral spastic cerebral palsy

BACKGROUND

Ground reaction forces are the gold standard for detecting gait events, but they are not always applicable in cerebral palsy. Ghoussayni's algorithm is an event detection method based on the sagittal plane velocity of heel and toe markers. We aimed to evaluate whether Ghoussayni's algorithm, using two different thresholds, was a valid event detection method in children with bilateral spastic cerebral palsy. We also aimed to define a new adaptation of Ghoussayni's algorithm for detecting foot strike in cerebral palsy, and study the effect of event detection methods on spatiotemporal parameters.

METHODS

Synchronized kinematic and kinetic data were collected retrospectively from 16 children with bilateral spastic cerebral palsy (7 males and 9 females; age 8.9 ± 2.7 years) walking barefoot at self-selected speed. Gait events were detected using methods: 1) ground reaction forces, 2) Ghoussayni's algorithm with a threshold of 0.5 m/s, and 3) Ghoussayni's algorithm with a walking speed dependent threshold. The new adaptation distinguished how foot strikes were performed (heel and/or toe) comparing the timing when the foot markers velocities fell below the threshold. Differences between the three methods, and between spatiotemporal parameters calculated from the two Ghoussayni's thresholds were analyzed.

FINDINGS

There were statistically significant (P < 0.05) differences between methods 1 and 3, and between some spatiotemporal parameters calculated from methods 2 and 3. Ghoussayni's algorithm showed better performance for foot strike than for toe off.

INTERPRETATION

Ghoussayni's algorithm using 0.5 m/s is valid in children with bilateral spastic cerebral palsy. Event detection methods affect spatiotemporal parameters.

Using a Deep Learning Method and Data from Two-Dimensional (2D) Marker-Less Video-Based Images for Walking Speed Classification

Human body measurement data related to walking can characterize functional movement and thereby become an important tool for health assessment. Single-camera-captured two-dimensional (2D) image sequences of marker-less walking individuals might be a simple approach for estimating human body measurement data which could be used in walking speed-related health assessment. Conventional body measurement data of 2D images are dependent on body-worn garments (used as segmental markers) and are susceptible to changes in the distance between the participant and camera in indoor and outdoor settings. In this study, we propose five ratio-based body measurement data that can be extracted from 2D images and can be used to classify three walking speeds (i.e., slow, normal, and fast) using a deep learning-based bidirectional long short-term memory classification model. The results showed that average classification accuracies of 88.08% and 79.18% could be achieved in indoor and outdoor environments, respectively. Additionally, the proposed ratio-based body measurement data are independent of body-worn garments and not susceptible to changes in the distance between the walking individual and camera. As a simple but efficient technique, the proposed walking speed classification has great potential to be employed in clinics and aged care homes.

Real-Time Musculoskeletal Kinematics and Dynamics Analysis Using Marker- and IMU-Based Solutions in Rehabilitation

This study aims to explore the possibility of estimating a multitude of kinematic and dynamic quantities using subject-specific musculoskeletal models in real-time. The framework was designed to operate with marker-based and inertial measurement units enabling extensions far beyond dedicated motion capture laboratories. We present the technical details for calculating the kinematics, generalized forces, muscle forces, joint reaction loads, and predicting ground reaction wrenches during walking. Emphasis was given to reduce computational latency while maintaining accuracy as compared to the offline counterpart. Notably, we highlight the influence of adequate filtering and differentiation under noisy conditions and its importance for consequent dynamic calculations. Real-time estimates of the joint moments, muscle forces, and reaction loads closely resemble OpenSim's offline analyses. Model-based estimation of ground reaction wrenches demonstrates that even a small error can negatively affect other estimated quantities. An application of the developed system is demonstrated in the context of rehabilitation and gait retraining. We expect that such a system will find numerous applications in laboratory settings and outdoor conditions with the advent of predicting or sensing environment interactions. Therefore, we hope that this open-source framework will be a significant milestone for solving this grand challenge.

Validation of IMU-based gait event detection during curved walking and turning in older adults and Parkinson's Disease patients

Background

Identification of individual gait events is essential for clinical gait analysis, because it can be used for diagnostic purposes or tracking disease progression in neurological diseases such as Parkinson’s disease. Previous research has shown that gait events can be detected from a shank-mounted inertial measurement unit (IMU), however detection performance was often evaluated only from straight-line walking. For use in daily life, the detection performance needs to be evaluated in curved walking and turning as well as in single-task and dual-task conditions.

Methods

Participants (older adults, people with Parkinson’s disease, or people who had suffered from a stroke) performed three different walking trials: (1) straight-line walking, (2) slalom walking, (3) Stroop-and-walk trial. An optical motion capture system was used a reference system. Markers were attached to the heel and toe regions of the shoe, and participants wore IMUs on the lateral sides of both shanks. The angular velocity of the shank IMUs was used to detect instances of initial foot contact (IC) and final foot contact (FC), which were compared to reference values obtained from the marker trajectories.

Results

The detection method showed high recall, precision and F1 scores in different populations for both initial contacts and final contacts during straight-line walking (IC: recall \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$=$$\end{document}= 100%, precision \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$=$$\end{document}= 100%, F1 score \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$=$$\end{document}= 100%; FC: recall \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$=$$\end{document}= 100%, precision \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$=$$\end{document}= 100%, F1 score \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$=$$\end{document}= 100%), slalom walking (IC: recall \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$=$$\end{document}= 100%, precision \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\ge$$\end{document}≥ 99%, F1 score \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$=$$\end{document}=100%; FC: recall \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$=$$\end{document}= 100%, precision \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\ge$$\end{document}≥ 99%, F1 score \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$=$$\end{document}=100%), and turning (IC: recall \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\ge$$\end{document}≥ 85%, precision \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\ge$$\end{document}≥ 95%, F1 score \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\ge$$\end{document}≥91%; FC: recall \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\ge$$\end{document}≥ 84%, precision \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\ge$$\end{document}≥ 95%, F1 score \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\ge$$\end{document}≥89%).

Conclusions

Shank-mounted IMUs can be used to detect gait events during straight-line walking, slalom walking and turning. However, more false events were observed during turning and more events were missed during turning. For use in daily life we recommend identifying turning before extracting temporal gait parameters from identified gait events.

Multidimensional Measures of Physical Activity and Their Association with Gross Motor Capacity in Children and Adolescents with Cerebral Palsy

The current lack of adapted performance metrics leads clinicians to focus on what children with cerebral palsy (CP) do in a clinical setting, despite the ongoing debate on whether capacity (what they do at best) adequately reflects performance (what they do in daily life). Our aim was to measure these children’s habitual physical activity (PA) and gross motor capacity and investigate their relationship. Using five synchronized inertial measurement units (IMU) and algorithms adapted to this population, we computed 22 PA states integrating the type (e.g., sitting, walking, etc.), duration, and intensity of PA. Their temporal sequence was visualized with a PA barcode from which information about pattern complexity and the time spent in each of the six simplified PA states (PAS; considering PA type and duration, but not intensity) was extracted and compared to capacity. Results of 25 children with CP showed no strong association between motor capacity and performance, but a certain level of motor capacity seems to be a prerequisite for the achievement of higher PAS. Our multidimensional performance measurement provides a new method of PA assessment in this population, with an easy-to-understand visual output (barcode) and objective data for clinical and scientific use.

Deep Learning in Gait Parameter Prediction for OA and TKA Patients Wearing IMU Sensors

Quantitative assessments of patient movement quality in osteoarthritis (OA), specifically spatiotemporal gait parameters (STGPs), can provide in-depth insight into gait patterns, activity types, and changes in mobility after total knee arthroplasty (TKA). A study was conducted to benchmark the ability of multiple deep neural network (DNN) architectures to predict 12 STGPs from inertial measurement unit (IMU) data and to identify an optimal sensor combination, which has yet to be studied for OA and TKA subjects. DNNs were trained using movement data from 29 subjects, walking at slow, normal, and fast paces and evaluated with cross-fold validation over the subjects. Optimal sensor locations were determined by comparing prediction accuracy with 15 IMU configurations (pelvis, thigh, shank, and feet). Percent error across the 12 STGPs ranged from 2.1% (stride time) to 73.7% (toe-out angle) and overall was more accurate in temporal parameters than spatial parameters. The most and least accurate sensor combinations were feet-thighs and singular pelvis, respectively. DNNs showed promising results in predicting STGPs for OA and TKA subjects based on signals from IMU sensors and overcomes the dependency on sensor locations that can hinder the design of patient monitoring systems for clinical application.

Design, Validity, and Reliability of a New Test, Based on an Inertial Measurement Unit System, for Measuring Cervical Posture and Motor Control in Children with Cerebral Palsy

OBJECTIVE

The aim of this study was to design and propose a new test based on inertial measurement unit (IMU) technology, for measuring cervical posture and motor control in children with cerebral palsy (CP) and to evaluate its validity and reliability.

METHODS

Twenty-four individuals with CP (4-14 years) and 24 gender- and age-matched controls were evaluated with a new test based on IMU technology to identify and measure any movement in the three spatial planes while the individual is seated watching a two-minute video. An ellipse was obtained encompassing 95% of the flexion/extension and rotation movements in the sagittal and transversal planes. The protocol was repeated on two occasions separated by 3 to 5 days. Construct and concurrent validity were assessed by determining the discriminant capacity of the new test and by identifying associations between functional measures and the new test outcomes. Relative reliability was determined using the intraclass correlation coefficient (ICC) for test-retest data. Absolute reliability was obtained by the standard error of measurement (SEM) and the Minimum Detectable Change at a 90% confidence level (MDC90).

RESULTS

The discriminant capacity of the area and both dimensions of the new test was high (Area Under the Curve ≈ 0.8), and consistent multiple regression models were identified to explain functional measures with new test results and sociodemographic data. A consistent trend of ICCs higher than 0.8 was identified for CP individuals. Finally, the SEM can be considered low in both groups, although the high variability among individuals determined some high MDC90 values, mainly in the CP group.

CONCLUSIONS

The new test, based on IMU data, is valid and reliable for evaluating posture and motor control in children with CP.

Walking Speed of Children and Adolescents With Cerebral Palsy: Laboratory Versus Daily Life

The purpose of this pilot study was to compare walking speed, an important component of gait, in the laboratory and daily life, in young individuals with cerebral palsy (CP) and with typical development (TD), and to quantify to what extent gait observed in clinical settings compares to gait in real life. Fifteen children, adolescents and young adults with CP (6 GMFCS I, 2 GMFCS II, and 7 GMFCS III) and 14 with TD were included. They wore 4 synchronized inertial sensors on their shanks and thighs while walking at their spontaneous self-selected speed in the laboratory, and then during 2 week-days and 1 weekend day in their daily environment. Walking speed was computed from shank angular velocity signals using a validated algorithm. The median of the speed distributions in the laboratory and daily life were compared at the group and individual levels using Wilcoxon tests and Spearman's correlation coefficients. The corresponding percentile of daily life speed equivalent to the speed in the laboratory was computed and observed at the group level. Daily-life walking speed was significantly lower compared to the laboratory for the CP group (0.91 [0.58-1.23] m/s vs 1.07 [0.73-1.28] m/s, p = 0.015), but not for TD (1.29 [1.24-1.40] m/s vs 1.29 [1.20-1.40] m/s, p = 0.715). Median speeds correlated highly in CP (p < 0.001, rho = 0.89), but not in TD. In children with CP, 60% of the daily life walking activity was at a slower speed than in-laboratory (corresponding percentile = 60). On the contrary, almost 60% of the daily life activity of TD was at a faster speed than in-laboratory (corresponding percentile = 42.5). Nevertheless, highly heterogeneous behaviors were observed within both populations and within subgroups of GMFCS level. At the group level, children with CP tend to under-perform during natural walking as compared to walking in a clinical environment. The heterogeneous behaviors at the individual level indicate that real-life gait performance cannot be directly inferred from in-laboratory capacity. This emphasizes the importance of completing clinical gait analysis with data from daily life, to better understand the overall function of children with CP.

Validation of a portable system for spatial-temporal gait parameters based on a single inertial measurement unit and a mobile application

There is a lack of commercially available low-cost technologies to assess gait clinically in non-controlled environments. As a consequence of this, there has been poor massification of motion measurement technologies that are both objective and reliable in nature. Advances about the study of gait and its interpretation in recent years using inertial sensors have allowed proposing acceptable alternatives for the development of portable and low-cost systems that contribute to people’s health in places and institutions that cannot acquire or maintain the operation of commercially available systems. A system based on a custom single Inertial Measurement Unit and a mobile application is proposed. Thus, an investigation is carried out using methodologies and algorithms found in the literature in order to get the main gait events and the spatial-temporal gait parameters. Twenty healthy Chilean subjects were assessed using a motion capture system simultaneously with the proposed tool. The results show that it is possible to estimate temporal gait parameters with slight differences respect gold--standard. We reach maximum mean differences of -2.35±5.02[step/min] for cadence, 0.03±0.04[sec] for stride time,0.02±0.03[sec] for step time, ±0.02[sec] for a single support time, 0.01±0.02[sec] for double support time and 0.01±0.03[m] for step length. As a result of experimental findings, we propose a new technological tool that can perform gait analysis. Our proposed system is user-friendly, low-cost, and portable. Therefore, we suggest that it could be an attractive technological tool that healthcare professionals could harness to objectively measure gait in environments that are either within the community or controlled. We also suggest that the tool could be used in countries where advanced clinical tools cannot be acquired. Therefore, we propose in this paper that our system is an attractive, alternative system that can be used for gait analysis by health professionals worldwide.

Self-perceived gait quality in young adults with cerebral palsy

AIM

To explore how patients with cerebral palsy (CP) perceive their gait and evaluate associations between subjective gait perception and: objective gait parameters, endurance, pain, and fatigue.

METHOD

Sixty-two patients (21 females and 41 males; mean [SD] age 20y [5y 1mo], range 15-29y) performed a clinical gait analysis. Self-selected walking speed, Gait Profile Score, and Gait Variable Score were calculated. Subjective gait perception was assessed with a visual analogue scale using the question: 'On a scale from 0 (worst) to 10 (optimal), how would you describe your walking today?'. A 6-minute walk test (6MWT) measured endurance; the 36-Item Short Form Health Survey (SF-36) evaluated quality of life. T-tests, Pearson correlations, and univariate and multiple linear regression models were used to compare and find associations between the data.

RESULTS

Overall mean (SD) subjective gait perception was 7.5 (1.8) and was significantly higher for patients in Gross Motor Function Classification System (GMFCS) level I (7.9 [1.5]) than for patients in GMFCS levels II and III (5.9 [2.0]). Positive correlations were found between subjective gait perception and gait scores, walking speed, 6MWT distance, and SF-36 score. Only walking speed was a significant predictor of subjective gait perception.

INTERPRETATION

Subjective gait perception was influenced by GMFCS level and linked partially with the walking speed. The gait quality did not explain subjective gait perception. It is important to combine subjective and objective gait scores to develop personalized therapeutic goals.

WHAT THIS PAPER ADDS

Subjective gait perception is influenced by the physical impairment levels of patients with cerebral palsy. Subjective gait perception and objective gait scores are associated. Walking speed is the only predictor of gait perception.

Validation of a portable system for spatial-temporal gait parameters based on a single inertial measurement unit and a mobile application

There is a lack of commercially available low-cost technologies to assess gait clinically in non-controlled environments. As a consequence of this, there has been poor massification of motion measurement technologies that are both objective and reliable in nature. Advances about the study of gait and its interpretation in recent years using inertial sensors have allowed proposing acceptable alternatives for the development of portable and low-cost systems that contribute to people’s health in places and institutions that cannot acquire or maintain the operation of commercially available systems. A system based on a custom single Inertial Measurement Unit and a mobile application is proposed. Thus, an investigation is carried out using methodologies and algorithms found in the literature in order to get the main gait events and the spatial-temporal gait parameters. Twenty healthy Chilean subjects were assessed using a motion capture system simultaneously with the proposed tool. The results show that it is possible to estimate temporal gait parameters with slight differences respect gold--standard. We reach maximum mean differences of -2.35±5.02[step/min] for cadence, 0.03±0.04[sec] for stride time,0.02±0.03[sec] for step time, ±0.02[sec] for a single support time, 0.01±0.02[sec] for double support time and 0.01±0.03[m] for step length. As a result of experimental findings, we propose a new technological tool that can perform gait analysis. Our proposed system is user-friendly, low-cost, and portable. Therefore, we suggest that it could be an attractive technological tool that healthcare professionals could harness to objectively measure gait in environments that are either within the community or controlled. We also suggest that the tool could be used in countries where advanced clinical tools cannot be acquired. Therefore, we propose in this paper that our system is an attractive, alternative system that can be used for gait analysis by health professionals worldwide.

Comparison of gait characteristics between clinical and daily life settings in children with cerebral palsy

Gait assessments in standardized settings, as part of the clinical follow-up of children with cerebral palsy (CP), may not represent gait in daily life. This study aimed at comparing gait characteristics in laboratory and real life settings on the basis of multiple parameters in children with CP and with typical development (TD). Fifteen children with CP and 14 with TD wore 5 inertial sensors (chest, thighs and shanks) during in-laboratory gait assessments and during 3 days of daily life. Sixteen parameters belonging to 8 distinct domains were computed from the angular velocities and/or accelerations. Each parameter measured in the laboratory was compared to the same parameter measured in daily life for walking bouts defined by a travelled distance similar to the laboratory, using Wilcoxon paired tests and Spearman’s correlations. Most gait characteristics differed between both environments in both groups. Numerous high correlations were found between laboratory and daily life gait parameters for the CP group, whereas fewer correlations were found in the TD group. These results demonstrated that children with CP perform better in clinical settings. Such quantitative evidence may enhance clinicians’ understanding of the gap between capacity and performance in children with CP and improve their decision-making.

Concurrent Validity and Reliability of an Inertial Measurement Unit for the Assessment of Craniocervical Range of Motion in Subjects with Cerebral Palsy

Objective: This study aimed to determine the validity and reliability of Inertial Measurement Units (IMUs) for the assessment of craniocervical range of motion (ROM) in patients with cerebral palsy (CP). Methods: twenty-three subjects with CP and 23 controls, aged between 4 and 14 years, were evaluated on two occasions, separated by 3 to 5 days. An IMU and a Cervical Range of Motion device (CROM) were used to assess craniocervical ROM in the three spatial planes. Validity was assessed by comparing IMU and CROM data using the Pearson correlation coefficient, the paired t-test and Bland–Altman plots. Intra-day and inter-day relative reliability were determined using the Intraclass Correlation Coefficient (ICC). The Standard Error of Measurement (SEM) and the Minimum Detectable Change at a 90% confidence level (MDC₉₀) were obtained for absolute reliability. Results: High correlations were detected between methods in both groups on the sagittal and frontal planes (r > 0.9), although this was reduced in the case of the transverse plane. Bland–Altman plots indicated bias below 5º, although for the range of cervical rotation in the CP group, this was 8.2º. The distance between the limits of agreement was over 23.5º in both groups, except for the range of flexion-extension in the control group. ICCs were higher than 0.8 for both comparisons and groups, except for inter-day comparisons of rotational range in the CP group. Absolute reliability showed high variability, with most SEM below 8.5º, although with worse inter-day results, mainly in CP subjects, with the MDC₉₀ of rotational range achieving more than 20º. Conclusions: IMU application is highly correlated with CROM for the assessment of craniocervical movement in CP and healthy subjects; however, both methods are not interchangeable. The IMU error of measurement can be considered clinically acceptable; however, caution should be taken when this is used as a reference measure for interventions.

A Personalized Approach to Improve Walking Detection in Real-Life Settings: Application to Children with Cerebral Palsy

Although many methods have been developed to detect walking by using body-worn inertial sensors, their performances decline when gait patterns become abnormal, as seen in children with cerebral palsy (CP). The aim of this study was to evaluate if fine-tuning an existing walking bouts (WB) detection algorithm by various thresholds, customized at the individual or group level, could improve WB detection in children with CP and typical development (TD). Twenty children (10 CP, 10 TD) wore 4 inertial sensors on their lower limbs during laboratory and out-laboratory assessments. Features extracted from the gyroscope signals recorded in the laboratory were used to tune thresholds of an existing walking detection algorithm for each participant (individual-based personalization: Indiv) or for each group (population-based customization: Pop). Out-of-laboratory recordings were analyzed for WB detection with three versions of the algorithm (i.e., original fixed thresholds and adapted thresholds based on the Indiv and Pop methods), and the results were compared against video reference data. The clinical impact was assessed by quantifying the effect of WB detection error on the estimated walking speed distribution. The two customized Indiv and Pop methods both improved WB detection (higher, sensitivity, accuracy and precision), with the individual-based personalization showing the best results. Comparison of walking speed distribution obtained with the best of the two methods showed a significant difference for 8 out of 20 participants. The personalized Indiv method excluded non-walking activities that were initially wrongly interpreted as extremely slow walking with the initial method using fixed thresholds. Customized methods, particularly individual-based personalization, appear more efficient to detect WB in daily-life settings.

Influence of BMI on Gait Characteristics of Young Adults: 3D Evaluation Using Inertial Sensors

Overweight/obesity is a physical condition that affects daily activities, including walking. The main purpose of this study was to identify if there is a relationship between body mass index (BMI) and gait characteristics in young adults. 12 normal weight (NW) and 10 overweight/obese (OW) individuals walked at a self-selected speed along a 14 m indoor path. H-Gait system, combining seven inertial sensors (fixed on pelvis and lower limbs), was used to record gait data. Walking speed, spatio-temporal parameters and joint kinematics in 3D were analyzed. Differences between NW and OW and correlations between BMI and gait parameters were evaluated. Conventional spatio-temporal parameters did not show statistical differences between the two groups or correlations with the BMI. However, significant results were pointed out for the joint kinematics. OW showed greater hip joint angles in frontal and transverse planes, with respect to NW. In the transverse plane, OW showed a greater knee opening angle and a shorter length of knee and ankle trajectories. Correlations were found between BMI and kinematic parameters in the frontal and transverse planes. Despite some phenomena such as soft tissue artifact and kinematics cross-talk, which have to be more deeply assessed, current results show a relationship between BMI and gait characteristics in young adults that should be looked at in osteoarthritis prevention.

Recurrent Neural Network for Inertial Gait User Recognition in Smartphones

In this article, a gait recognition algorithm is presented based on the information obtained from inertial sensors embedded in a smartphone, in particular, the accelerometers and gyroscopes typically embedded on them. The algorithm processes the signal by extracting gait cycles, which are then fed into a Recurrent Neural Network (RNN) to generate feature vectors. To optimize the accuracy of this algorithm, we apply a random grid hyperparameter selection process followed by a hand-tuning method to reach the final hyperparameter configuration. The different configurations are tested on a public database with 744 users and compared with other algorithms that were previously tested on the same database. After reaching the best-performing configuration for our algorithm, we obtain an equal error rate (EER) of 11.48% when training with only 20% of the users. Even better, when using 70% of the users for training, that value drops to 7.55%. The system manages to improve on state-of-the-art methods, but we believe the algorithm could reach a significantly better performance if it was trained with more visits per user. With a large enough database with several visits per user, the algorithm could improve substantially.

Application of an Improved Correlation Method in Electrostatic Gait Recognition of Hemiparetic Patients

Hemiparesis is one of the common sequelae of neurological diseases such as strokes, which can significantly change the gait behavior of patients and restrict their activities in daily life. The results of gait characteristic analysis can provide a reference for disease diagnosis and rehabilitation; however, gait correlation as a gait characteristic is less utilized currently. In this study, a new non-contact electrostatic field sensing method was used to obtain the electrostatic gait signals of hemiplegic patients and healthy control subjects, and an improved Detrended Cross-Correlation Analysis cross-correlation coefficient method was proposed to analyze the obtained electrostatic gait signals. The results show that the improved method can better obtain the dynamic changes of the scaling index under the multi-scale structure, which makes up for the shortcomings of the traditional Detrended Cross-Correlation Analysis cross-correlation coefficient method when calculating the electrostatic gait signal of the same kind of subjects, such as random and incomplete similarity in the trend of the scaling index spectrum change. At the same time, it can effectively quantify the correlation of electrostatic gait signals in subjects. The proposed method has the potential to be a powerful tool for extracting the gait correlation features and identifying the electrostatic gait of hemiplegic patients.

Real-World Gait Speed Estimation Using Wrist Sensor: A Personalized Approach

Gait speed is an important parameter to characterize people's daily mobility. For real-world speed measurement, inertial sensors or global navigation satellite system (GNSS) can be used on wrist, possibly integrated in a wristwatch. However, power consumption of GNSS is high and data are only available outdoor. Gait speed estimation using wrist-mounted inertial sensors is generally based on machine learning and suffers from low accuracy because of the inadequacy of using limited training data to build a general speed model that would be accurate for the whole population. To overcome this issue, a personalized model was proposed, which took unique gait style of each subject into account. Cadence and other biomechanically derived gait features were extracted from a wrist-mounted accelerometer and barometer. Gait features were fused with few GNSS data (sporadically sampled during gait) to calibrate the step length model of each subject through online learning. The proposed method was validated on 30 healthy subjects where it has achieved a median [Interquartile Range] of root mean square error of 0.05 [0.04-0.06] (m/s) and 0.14 [0.11-0.17] (m/s) for walking and running, respectively. Results demonstrated that the personalized model provided similar performance as GNSS. It used 50 times less training GNSS data than nonpersonalized method and achieved even better results. This parsimonious GNSS usage allowed extending battery life. The proposed algorithm met requirements for applications which need accurate, long, real-time, low-power, and indoor/outdoor speed estimation in daily life.

Using smartphone accelerometry to assess the relationship between cognitive load and gait dynamics during outdoor walking

Research has demonstrated that an increase in cognitive load can result in increased gait variability and slower overall walking speed, both of which are indicators of gait instability. The external environment also imposes load on our cognitive systems; however, most gait research has been conducted in a laboratory setting and little work has demonstrated how load imposed by natural environments impact gait dynamics during outdoor walking. Across four experiments, young adults were exposed to varying levels of cognitive load while walking through indoor and outdoor environments. Gait dynamics were concurrently recorded using smartphone-based accelerometry. Results suggest that, during indoor walking, increased cognitive load impacted a range of gait parameters such as step time and step time variability. The impact of environmental load on gait, however, was not as pronounced, with increased load associated only with step time changes during outdoor walking. Overall, the present work shows that cognitive load is related to young adult gait during both indoor and outdoor walking, and importantly, smartphones can be used as gait assessment tools in environments where gait dynamics have traditionally been difficult to measure.

Locomotion and cadence detection using a single trunk-fixed accelerometer: validity for children with cerebral palsy in daily life-like conditions

BACKGROUND

Physical therapy interventions for ambulatory youth with cerebral palsy (CP) often focus on activity-based strategies to promote functional mobility and participation in physical activity. The use of activity monitors validated for this population could help to design effective personalized interventions by providing reliable outcome measures. The objective of this study was to devise a single-sensor based algorithm for locomotion and cadence detection, robust to atypical gait patterns of children with CP in the real-life like monitoring conditions.

METHODS

Study included 15 children with CP, classified according to Gross Motor Function Classification System (GMFCS) between levels I and III, and 11 age-matched typically developing (TD). Six IMU devices were fixed on participant's trunk (chest and low back/L5), thighs, and shanks. IMUs on trunk were independently used for development of algorithm, whereas the ensemble of devices on lower limbs were used as reference system. Data was collected according to a semi-structured protocol, and included typical daily-life activities performed indoor and outdoor. The algorithm was based on detection of peaks associated to heel-strike events, identified from the norm of trunk acceleration signals, and included several processing stages such as peak enhancement and selection of the steps-related peaks using heuristic decision rules. Cadence was estimated using time- and frequency-domain approaches. Performance metrics were sensitivity, specificity, precision, error, intra-class correlation coefficient, and Bland-Altman analysis.

RESULTS

According to GMFCS, CP children were classified as GMFCS I (n = 7), GMFCS II (n = 3) and GMFCS III (n = 5). Mean values of sensitivity, specificity and precision for locomotion detection ranged between 0.93-0.98, 0.92-0.97 and 0.86-0.98 for TD, CP-GMFCS I and CP-GMFCS II-III groups, respectively. Mean values of absolute error for cadence estimation (steps/min) were similar for both methods, and ranged between 0.51-0.88, 1.18-1.33 and 1.94-2.3 for TD, CP-GMFCS I and CP-GMFCS II-III groups, respectively. The standard deviation was higher in CP-GMFCS II-III group, the lower performances being explained by the high variability of atypical gait patterns.

CONCLUSIONS

The algorithm demonstrated good performance when applied to a wide range of gait patterns, from normal to the pathological gait of highly affected children with CP using walking aids.

A 'Fingerprint' of locomotor maturation: Motor development descriptors, reference development bands and data-set

BACKGROUND

When aiming at studying and monitoring locomotor development in childhood, innovative indexes for the characterization of motor control performance and wearable technologies have highlighted the potential of significant advances. In particular, quantitative assessment of motor performance during natural walking (NW) and tandem walking (TW) has been proposed to highlight manifestations of motor automaticity and complexity, respectively.

RESEARCH QUESTION

This work aims at providing a quantitative overview of metrics characterizing locomotor maturation in a typically developing population, by analysing NW and TW. The final goal is to propose a novel graphical representation of motor development from childhood to adulthood, providing metrics for quantitative assessment with reference bands and data-set, supporting data interpretation and longitudinal assessment.

METHODS

112 typically developing participants (age groups: 6-, 7-, 8-, 9-, 10-, 15-, and 25 years) walked in NW and in TW at self-selected speed. 3D acceleration and angular velocity of lower trunk and shanks were collected. Temporal parameters, their variability, and nonlinear metrics characterizing human movement (harmonic ratio, short-term Lyapunov exponents, multiscale entropy, and recurrence quantification analysis) were calculated. Effect of age was analysed on the different parameters and a graphical polar plot was defined to represent parameters that showed age effect in at least one of the two tasks.

RESULTS

Age effect was shown on temporal parameters, their variability, multiscale entropy and recurrence quantification analysis. These parameters were selected for monitoring locomotor development and presented on an ad-hoc designed polar plot showing age-group reference bands.

SIGNIFICANCE

Graphic results outline locomotor differences with maturation at first glance. The patterns in NW and TW allow to characterize specific aspects of locomotor maturation, to evaluate in which area changes occur and towards which direction, depending on the task. The novel database containing participants' raw collected data is made available as additional result of the present study.

Weight-Bearing Estimation for Cane Users by Using Onboard Sensors

Mobility is a fundamental requirement for a healthy, active lifestyle. Gait analysis is widely acknowledged as a clinically useful tool for identifying problems with mobility, as identifying abnormalities within the gait profile is essential to correct them via training, drugs, or surgical intervention. However, continuous gait analysis is difficult to achieve due to technical limitations, namely the need for specific hardware and constraints on time and test environment to acquire reliable data. Wearables may provide a solution if users carry them most of the time they are walking. We propose to add sensors to walking canes to assess user's mobility. Canes are frequently used by people who cannot completely support their own weight due to pain or balance issues. Furthermore, in absence of neurological disorders, the load on the cane is correlated with the user condition. Sensorized canes already exist, but often rely on expensive sensors and major device modifications are required. Thus, the number of potential users is severely limited. In this work, we propose an affordable module for load monitoring so that it can be widely used as a screening tool. The main advantages of our module are: (i) it can be deployed in any standard cane with minimal changes that do not affect ergonomics; (ii) it can be used every day, anywhere for long-term monitoring. We have validated our prototype with 10 different elderly volunteers that required a cane to walk, either for balance or partial weight bearing. Volunteers were asked to complete a 10 m test and, then, to move freely for an extra minute. The load peaks on the cane, corresponding to maximum support instants during the gait cycle, were measured while they moved. For validation, we calculated their gait speed using a chronometer during the 10 m test, as it is reportedly related to their condition. The correlation between speed (condition) and load results proves that our module provides meaningful information for screening. In conclusion, our module monitors support in a continuous, unsupervised, nonintrusive way during users' daily routines, plus only mechanical adjustment (cane height) is needed to change from one user to another.

Multiple-Wearable-Sensor-Based Gait Classification and Analysis in Patients with Neurological Disorders

The aim of this study was to conduct a comprehensive analysis of the placement of multiple wearable sensors for the purpose of analyzing and classifying the gaits of patients with neurological disorders. Seven inertial measurement unit (IMU) sensors were placed at seven locations: the lower back (L5) and both sides of the thigh, distal tibia (shank), and foot. The 20 subjects selected to participate in this study were separated into two groups: stroke patients (11) and patients with neurological disorders other than stroke (brain concussion, spinal injury, or brain hemorrhage) (9). The temporal parameters of gait were calculated using a wearable device, and various features and sensor configurations were examined to establish the ideal accuracy for classifying different groups. A comparison of the various methods and features for classifying the three groups revealed that a combination of time domain and gait temporal feature-based classification with the Multilayer Perceptron (MLP) algorithm outperformed the other methods of feature-based classification. The classification results of different sensor placements revealed that the sensor placed on the shank achieved higher accuracy than the other sensor placements (L5, foot, and thigh). The placement-based classification of the shank sensor achieved 89.13% testing accuracy with the Decision Tree (DT) classifier algorithm. The results of this study indicate that the wearable IMU device is capable of differentiating between the gait patterns of healthy patients, patients with stroke, and patients with other neurological disorders. Moreover, the most favorable results were reported for the classification that used the combination of time domain and gait temporal features as the model input and the shank location for sensor placement.