Comparing Gait Trials with Greedy Template Matching

Preventing falls: the use of machine learning for the prediction of future falls in individuals without history of fall

Nowadays, it becomes of paramount societal importance to support many frail-prone groups in our society (elderly, patients with neurodegenerative diseases, etc.) to remain socially and physically active, maintain their quality of life, and avoid their loss of autonomy. Once older people enter the prefrail stage, they are already likely to experience falls whose consequences may accelerate the deterioration of their quality of life (injuries, fear of falling, reduction of physical activity). In that context, detecting frailty and high risk of fall at an early stage is the first line of defense against the detrimental consequences of fall. The second line of defense would be to develop original protocols to detect future fallers before any fall occur. This paper briefly summarizes the current advancements and perspectives that may arise from the combination of affordable and easy-to-use non-wearable systems (force platforms, 3D tracking motion systems), wearable systems (accelerometers, gyroscopes, inertial measurement units-IMUs) with appropriate machine learning analytics, as well as the efforts to address these challenges.

A topological data analysis-based method for gait signals with an application to the study of multiple sclerosis

In the past few years, light, affordable wearable inertial measurement units have been providing to clinicians and researchers the possibility to quantitatively study motor degeneracy by comparing gait trials from patients and/or healthy subjects. To do so, standard gait features can be used but they fail to detect subtle changes in several pathologies including multiple sclerosis. Multiple sclerosis is a demyelinating disease of the central nervous system whose symptoms include lower limb impairment, which is why gait trials are commonly used by clinicians for their patients’ follow-up. This article describes a method to compare pairs of gait signals, visualize the results and interpret them, based on topological data analysis techniques. Our method is non-parametric and requires no data other than gait signals acquired with inertial measurement units. We introduce tools from topological data analysis (sublevel sets, persistence barcodes) in a practical way to make it as accessible as possible in order to encourage its use by clinicians. We apply our method to study a cohort of patients suffering from progressive multiple sclerosis and healthy subjects. We show that it can help estimate the severity of the disease and also be used for longitudinal follow-up to detect an evolution of the disease or other phenomena such as asymmetry or outliers.

The Diverse Gait Dataset: Gait Segmentation Using Inertial Sensors for Pedestrian Localization with Different Genders, Heights and Walking Speeds

Stride length estimation is one of the most crucial aspects of Pedestrian Dead Reckoning (PDR). Due to the measurement noise of inertial sensors, individual variances of pedestrians, and the uncertainty in pedestrians walking, there is a substantial error in the assessment of stride length, which causes the accumulated deviation of Pedestrian Dead Reckoning (PDR). With the help of multi-gait analysis, which decomposes strides in time and space with greater detail and accuracy, a novel and revolutionary stride estimating model or scheme could improve the performance of PDR on different users. This paper presents a diverse stride gait dataset by using inertial sensors that collect foot movement data from people of different genders, heights, and walking speeds. The dataset contains 4690 walking strides data and 19,083 gait labels. Based on the dataset, we propose a threshold-independent stride segmentation algorithm called SDATW and achieve an F-measure of 0.835. We also provide the detailed results of recognizing four gaits under different walking speeds, demonstrating the utility of our dataset for helping train stride segmentation algorithms and gait detection algorithms.

Postadychute-AG, Detection, and Prevention of the Risk of Falling Among Elderly People in Nursing Homes: Protocol of a Multicentre and Prospective Intervention Study

Introduction: While falls among the elderly is a public health issue, because of the social, medical, and economic burden they represent, the tools to predict falls are limited. Posturography has been developed to distinguish fallers from non-fallers, however, there is too little data to show how predictions change as older adults' physical abilities improve. The Postadychute-AG clinical trial aims to evaluate the evolution of posturographic parameters in relation to the improvement of balance through adapted physical activity (APA) programs. Methods: In this prospective, multicentre clinical trial, institutionalized seniors over 65 years of age will be followed for a period of 6 months through computer-assisted posturography and automatic gait analysis. During the entire duration of the follow-up, they will benefit from a monthly measurement of their postural and locomotion capacities through a recording of their static balance and gait thanks to a software developed for this purpose. The data gathered will be correlated with the daily record of falls in the institution. Static and dynamic balance measurements aim to extract biomechanical markers and compare them with functional assessments of motor skills (Berg Balance Scale and Mini Motor Test), expecting their superiority in predicting the number of falls. Participants will be followed for 3 months without APA and 3 months with APA in homogeneous group exercises. An analysis of variance will evaluate the variability of monthly measures of balance in order to record the minimum clinically detectable change (MDC) as participants improve their physical condition through APA. Discussion: Previous studies have stated the MDC through repeated measurements of balance but, to our knowledge, none appear to have implemented monthly measurements of balance and gait. Combined with a reliable measure of the number of falls per person, motor capacities and other precipitating factors, this study aims to provide biomechanical markers predictive of fall risk with their sensitivity to improvement in clinical status over the medium term. This trial could provide the basis for posturographic and gait variable values for these elderly people and provide a solution to distinguish those most at risk to be implemented in current practice in nursing homes. Trial Registration: ID-RCB 2017-A02545-48. Protocol Version: Version 4.2 dated January 8, 2020.

Personalized Template-Based Step Detection From Inertial Measurement Units Signals in Multiple Sclerosis

Background: Objective gait assessment is key for the follow-up of patients with progressive multiple sclerosis (pMS). Inertial measurement units (IMUs) provide reliable and yet easy quantitative gait assessment in routine clinical settings. However, to the best of our knowledge, no automated step-detection algorithm performs well in detecting severely altered pMS gait. Method: This article elaborates on a step-detection method based on personalized templates tested against a gold standard. Twenty-two individuals with pMS and 10 young healthy subjects (HSs) were instructed to walk on an electronic walkway wearing synchronized IMUs. Templates were derived from the IMU signals by using Initial and Final Contact times given by the walkway. These were used to detect steps from other gait trials of the same individual (intra-individual template-based detection, IITD) or another participant from the same group (pMS or HS) (intra-group template-based detection, IGTD). All participants were seen twice with a 6-month interval, with two measurements performed at each visit. Performance and accuracy metrics were computed, along with a similarity index (SId), which was computed as the mean distance between detected steps and their respective closest template. Results: For HS participants, both the IITD and the IGTD algorithms had precision and recall of 1.00 for detecting steps. For pMS participants, precision and recall ranged from 0.94 to 1.00 for IITD and 0.85 to 0.95 for IGTD depending on the level of disability. The SId was correlated with performance and the accuracy of the result. An SId threshold of 0.957 (IITD) and 0.963 (IGTD) could rule out decreased performance (F-measure ≤ 0.95), with negative predictive values of 0.99 and 0.96 with the IITD and IGTD algorithms. Also, the SId computed with the IITD and IGTD algorithms could distinguish individuals showing changes at 6-month follow-up. Conclusion: This personalized step-detection method has high performance for detecting steps in pMS individuals with severely altered gait. The algorithm can be self-evaluating with the SI, which gives a measure of the confidence the clinician can have in the detection. What is more, the SId can be used as a biomarker of change in disease severity occurring between the two measurement times.

Non-Linear Template-Based Approach for the Study of Locomotion

The automatic detection of gait events (i.e., Initial Contact (IC) and Final Contact (FC)) is crucial for the characterisation of gait from Inertial Measurements Units. In this article, we present a method for detecting steps (i.e., IC and FC) from signals of gait sequences of individuals recorded with a gyrometer. The proposed approach combines the use of a dictionary of templates and a Dynamic Time Warping (DTW) measure of fit to retrieve these templates into input signals. Several strategies for choosing and learning the adequate templates from annotated data are also described. The method is tested on thirteen healthy subjects and compared to gold standard. Depending of the template choice, the proposed algorithm achieves average errors from 0.01 to 0.03 s for the detection of IC, FC and step duration. Results demonstrate that the use of DTW allows achieving these performances with only one single template. DTW is a convenient tool to perform pattern recognition on gait gyrometer signals. This study paves the way for new step detection methods: it shows that using one single template associated with non-linear deformations may be sufficient to model the gait of healthy subjects.