The human ankle joint is hypothesized to be a primary controller of support, propulsion and steering during locomotion. A series of experiments were initiated to understand ankle plantarflexor muscle kinematics and kinetics in normal and pathological gait, and to define the specific locomotor demands of community ambulation. Additional experiments were then conducted to quantify the effects of walking speed on plantar pressures and centre of mass motion, to illuminate the role of the ankle in acceleration and deceleration during walking, and to examine how humans alter their kinematics and kinetics to turn. The results of these experiments provide support for the hypothesis that the ankle joint is important in a wide range of locomotor movements beyond walking straight ahead at constant speed. The ankle appears instrumental in adapting to different walking speeds, altering both the pressures on specific regions the plantar surface and the motion of the centre of mass across a range of speeds. The ankle also has subtle kinetic changes that appear to modulate acceleration and deceleration during single limb stance. For turning, the ankle plays a role during slowing into the turn and accelerating after the turn, but mediolateral shears appear to alter the trajectory of the body to negotiate a corner and the external hip rotators appear to rotate the trunk toward the new direction of travel. This work extends our understanding of the ankle in functionally relevant gait activities beyond simple straight-ahead walking at constant speed. The published papers included in this supporting statement have been cited by 180 different subsequent peerreviewed publications, suggesting that this work has had some impact on the field.

Human activity analysis has attracted great interest from computer vision researchers due to its promising applications in many areas such as automated visual surveillance, computer-human interactions, and motion-based identification and diagnosis. This dissertation presents work in two areas: general human activity recognition from video, and human activity analysis for the purpose of identifying pathological gait from both 3D captured data and from video. Even though the research in human activity recognition has been going on for many years, still there are many issues that need more research. This includes the effective representation and modeling of human activities and the segmentation of sequences of continuous activities. In this thesis we present an algorithm that combines shape and motion features to represent human activities. In order to handle the activity recognition from any viewing angle we quantize the viewing direction and build a set of Hidden Markov Models (HMMs), where each model represents the activity from a given view. Finally, a voting based algorithm is used to segment and recognize a sequence of human activities from video. Our method of representing activities has good attributes and is suitable for both low resolution and high resolution video. The voting based algorithm performs the segmentation and recognition simultaneously. Experiments on two sets of video clips of different activities show that our method is effective. Our work on identifying pathological gait is based on the assumption of gait symmetry. Previous work on gait analysis measures the symmetry of gait based on Ground Reaction Force data, stance time, swing time or step length. Since the trajectories of the body parts contain information about the whole body movement, we measure the symmetry of the gait based on the trajectories of the body parts. Two algorithms, which can work with different data sources, are presented. The first algorithm works on 3D motion-captured data and the second works on video data. Both algorithms use support vector machine (SVM) for classification. Each of the two methods has three steps: the first step is data preparation, i.e., obtaining the trajectories of the body parts; the second step is gait representation based on a measure of gait symmetry; and the last step is SVM based classification. For 3D motion-captured data, a set of features based on Discrete Fourier Transform (DFT) is used to represent the gait. We demonstrate the accuracy of the classification by a set of experiments that shows that the method for 3D motion-captured data is highly effective. For video data, a model based tracking algorithm for human body parts is developed for preparing the data. Then, a symmetry measure that works on the sequence of 2D data, i.e. sequence of video frames, is derived to represent the gait. We performed experiments on both 2D projected data and real video data to examine this algorithm. The experimental results on 2D projected data showed that the presented algorithm is promising for identifying pathological gait from video. The experimental results on the real video data are not good as the results on 2D projected data. We believe that better results could be obtained if the accuracy of the tracking algorithm is improved.

L’objectif principal de ce travail est d’étudier la faisabilité de réaliser des mesures du travail mécanique au cours de différentes marches pathologiques et de répondre à la question suivante : « La mesure du travail mécanique permet-elle d’expliquer l’augmentation du coût énergétique de la marche dans différents déficits unilatéraux ? ». L’énergétique et la mécanique de la marche ont été évaluées dans les quatre déficits unilatéraux suivants : l’hémiparésie, la gonarthrose unilatérale, l’arthroplastie totale d’un genou et l’amputation partielle d’un membre inférieur. Au terme de ce travail, il apparaît que : (a) les mesures du travail mécanique sont réalisables dans les quatre déficits unilatéraux choisis ; (b) les variables mécaniques permettent d’expliquer l’augmentation du coût énergétique de la marche ; (c) l’augmentation du coût énergétique ne peut être expliquée par l’inefficacité du mécanisme pendulaire de la marche.

La marche est le mode de locomotion naturel de l’homme. Malgré les très nombreuses études s’y étant intéressé, cela reste un mouvement complexe. Ceci est d’autant plus vrai lorsqu’une pathologie vient le perturber. Dans le cadre clinique, le recueil de données réalisé est appelé l’Analyse Quantifiée de la Marche (AQM). Elle s’adresse, notamment, à des patients souffrant de troubles de la marche issus de pathologies affectant le système nerveux central. La quantité dedonnées pouvant être extraite d’une AQM étant très importante, des index ont été définis et validés. Le Gillette Gait Index (GGI), le Gait Deviation Index (GDI) et l’Edinburgh Visual Gait Score (EVGS) sont parmi les plus utilisés. Leurs limites principales sont qu’ils ont été définis que pour la prise en charge des enfants paralysés cérébraux et qu’ils sont basés presque exclusivement sur la cinématique. Les modalités de calcul de ces index n’étant pas spécifiques de la pédiatrie, dans un premier temps, nous avons voulu voir comment ceux-ci se comportaient chez l’adulte. Nous avons ainsi, par la démonstration de l’applicabilité du GGI, du GDI et de l’EVGS, validé le principe de l’AQM chez l’adulte. Cependant, de façon courante, l’AQM comprend dans son protocole un enregistrement électromyographique (EMG) qui ne fait que très rarement l’objet d’une réelle quantification à la marche. Nous avons donc, dans un second temps, défini un nouvel index dequantification de la marche pathologique basé sur l’EMG : le Ker-EGI. Cet index reprend la philosophie et le modèle mathématique du GDI. Nous avons validé le Ker-EGI chez l’adulte en le corrélant avec le GGI, le GDI et l’EVGS. Ce nouvel index va permettre de réaliser, à moindre coût, un meilleur suivi au quotidien des patients. Il est plus accessible en routine clinique et pourra être associé à l’EVGS pour donner un tableau clinique complet du patient (regards neuromoteur etcinématique) / Walking is the natural way of locomotion for human. Nevertheless, despite numerous studies, it remains a complex movement. This is all the more real when a pathology disturbs it. Data collection made on patients is called Clinical Gait Analysis (CGA). This is dedicated, in particular, to patients with a central nervous system disorder. As data outcoming from the CGA could be very heavy, indices have been defined and validated. Among the most used are the Gillette Gait Index(GGI), the Gait Deviation Index (GDI) and the Edinburgh Visual Gait Score (EVGS). Their main limitations are that they have only been defined for children with cerebral palsy and they are based quite solely one kinematics. As the methods to compute these indices are not child-specific, we have first evaluated how they could also be used in adults. So, demonstrating the applicability of the GGI, the GDI and the EVGS, we have validated the principle of CGA in adults. Usually, the CGA’s protocol includes electromyographic measures (EMG), but rarely these data are really quantified. That is why, secondly, we have defined a new index of gait quantification based on EMG: the Ker-EGI. This index uses the philosophy and the mathematical model of the GDI. We have validated the Ker-EGI in adults correlating it with the GGI, the GDI and the EVGS. This new index more accessible in clinical routine allows to perform, for a lower cost, a better patient’s care in everyday life. Furthermore, if the Ker-EGI is associated with the EVGS, we have a more complete clinical picture with a neuro-motor and kinematic view of the patient

Les systèmes de biofeedback sont encore développés de façon ad 'hoc de nos jours. Aucun modèle ou cadre conceptuel n'existe qui regroupe et synthétise l'ensemble des découvertes réalisées, cela rendant difficile, voire impossible la comparaison d'un système à un autre pas plus que la réutilisation d'une ou plusieurs parties. Peu de dispositifs impliquent l'utilisateur dès les premières étapes de la conception. Il est pourtant partie prenante, dans ses attentes envers le système et dans la prise en compte de ses capacités physiologiques ou cognitives. L'information qui lui est restituée aura un impact certain, tant sur son engagement personnel dans le processus de rééducation que dans sa motivation à le mener jusqu'à son terme. Dans ce contexte, l'objectif de ces travaux de thèse est de faire un apport dans la conception et le développement des dispositifs de biofeedback, en adoptant une approche qui considère le patient avant tout comme l'utilisateur d'un système et en le plaçant au centre la boucle de biofeedback. Après deux revues de littérature sur les systèmes de biofeedback et sur l'utilisation de la modalité audio, nous présentons notre première contribution : le cadre de conception de systèmes de biofeedback centré utilisateur qui synthétise les meilleures pratiques de la littérature à travers une terminologie commune. Il met en évidence les principaux composants qui interviennent dans la boucle de biofeedback : les capteurs, les données, les traitements de bas niveau, la fusion multi capteurs, l'exploitation de haut niveau et enfin le rendu. Nous complétons ce cadre conceptuel par ce que nous considérons comme un point fondamental : la chaîne d'influence. Elle permet de définir ce qui a trait à l'étude des systèmes de biofeedback, mais aide aussi à préciser clairement les buts des parties prenantes que sont le praticien et l'utilisateur, les capacités de ce dernier et enfin le contexte d'utilisation. Les contributions suivantes mettent l'accent sur l'information présentée en retour à l'utilisateur : le rendu. Une première expérimentation évalue les niveaux sonores discriminables par les utilisateurs sans apprentissage. La seconde compare différentes techniques décrivant l'évolution des pressions plantaires dans le cadre de la marche afin d'évaluer la technique la plus à même de permettre aux utilisateurs de discriminer une marche pathologique d'une marche saine. Nous présentons enfin les outils d'exploration sonore que nous avons conçus et développés, complétés par un émulateur de dispositif de pressions plantaires. / Biofeedback devices are still ad 'hoc developing nowadays. No model or conceptual framework exists which gathers and summarizes all the discoveries, making it difficult or impossible to compare one system to another nor reuse of one or more element. Few devices involve the user in the early stages of design. However, he is involved in its expectations of the system and in consideration of its physiological and cognitive abilities. The information that is returned to him will have an impact on both his personal commitment to the rehabilitation process in its motivation to carry on to completion. In this context, the purpose of this thesis work is to make a contribution in the design and development of biofeedback devices, adopting an approach that considers the patient foremost as the system's user and placing him in the centre of the biofeedback loop. Following two literature reviews on biofeedback systems and the use of the auditory modality, we present our first contribution: a user centered biofeedback system design framework that synthesizes the best practices of literature through a common terminology. It highlights the main components involved in the biofeedback loop: sensors, data, low-level processing, multi-sensor fusion, high-level operation and finally rendering. We complete this framework by what we see as a fundamental point: the chain of influence. It relates to the study of biofeedback systems, but also helps to clearly define the goals of the stakeholders that are the practitioner and the user. It cares for the capabilities of the latter and finally the context of use. The following contributions focus on the information presented back to the user: the rendering. A first experiment evaluates discriminable sound levels by users without learning. The second compares different techniques describing the evolution of plantar pressure within walking to assess the most likely way to allow users to distinguish a pathological walking of healthy walking. Finally, we present the sound exploration tools we have designed and developed, complemented by a plantar pressure device emulator.

Spinal cord injury, whether traumatic or nontraumatic, can partially or completely damage sensorimotor pathways between the brain and the body, leading to heterogeneous gait abnormalities. Mobility impairments also depend on other factors such as age, weight, time since injury, pain, and walking aids used. The ASIA Impairment Scale is recommended to classify injury severity, but is not designed to characterize individual ambulatory capacity. Other standardized tests based on subjective or timing/distance assessments also have only limited ability to determine an individual's capacity. Data-driven techniques have demonstrated effectiveness in analysing complexity in many domains and may provide additional perspectives on the complexity of gait performance in persons with spinal cord injury. The studies in this thesis aimed to address the complexity of gait and functional abilities after spinal cord injury using data-driven approaches. The aim of the first manuscript was to characterize the heterogeneous gait patterns in persons with incomplete spinal cord injury. Dissimilarities among gait patterns in the study population were quantified with multivariate dynamic time warping. Gait patterns were classified into six distinct clusters using hierarchical agglomerative clustering. Through random forest classifiers with explainable AI, peak ankle plantarflexion during swing was identified as the feature that most often distinguished most clusters from the controls. By combining clinical evaluation with the proposed methods, it was possible to provide comprehensive analyses of the six gait clusters.     The aim of the second manuscript was to quantify sensorimotor effects on walking performance in persons with spinal cord injury. The relationships between 11 input features and 2 walking outcome measures - distance walked in 6 minutes and net energy cost of transport - were captured using 2 Gaussian process regression models. Explainable AI revealed the importance of muscle strength on both outcome measures. Use of walking aids also influenced distance walked, and  cardiovascular capacity influenced energy cost. Analyses for each person also gave useful insights into individual performance.     The findings from these studies demonstrate the large potential of advanced machine learning and explainable AI to address the complexity of gait function in persons with spinal cord injury. / Skador på ryggmärgen, oavsett om de är traumatiska eller icke-traumatiska, kan helt eller delvis skada sensoriska och motoriska banor mellan hjärnan och kroppen, vilket påverkar gången i varierande grad. Rörelsenedsättningen beror också på andra faktorer såsom ålder, vikt, tid sedan skadan uppstod, smärta och gånghjälpmedel. ASIA-skalan används för att klassificera ryggmärgsskadans svårighetsgrad, men är inte utformad för att karaktärisera individens gångförmåga. Andra standardiserade tester baserade på subjektiva eller tids och avståndsbedömningar har också begränsad möjlighet att beskriva individuell kapacitet. Datadrivna metoder är kraftfulla och kan ge ytterligare perspektiv på gångens komplexitet och prestation. Studierna i denna avhandling syftar till att analysera komplexa relationer mellan gång, motoriska samt sensoriska funktion efter ryggmärgsskada med hjälp av datadrivna metoder. Syftet med den första studien är att karaktärisera de heterogena gångmönster hos personer med inkomplett ryggmärgsskada. Multivariat dynamisk tidsförvrägning (eng: Multivariate dynamic time warping) användes för att kvantifiera gångskillnader i studiepopulationen. Hierarkisk agglomerativ klusteranalys (eng: hierarchical agglomerative clustering) delade upp gång i sex distinkta kluster, varav fyra hade lägre hastighet än kontroller. Med hjälp av förklarbara AI (eng: explainable AI) identifierades det att fotledsvinkeln i svingfasen hade störst påverkan om vilken kluster som gångmönstret hamnat i. Genom att kombinera klinisk undersökning med datadrivna metoder kunde vi beskriva en omfattande bild av de sex gångklustren. Syftet med den andra manuskriptet är att kvantifiera sensoriska och motoriska faktorerans påverkan på gångförmåga efter ryggmärgsskada. Med hjälp av två Gaussian process-regressionsmodeller identiferades sambanden mellan 11 beskrivande faktorer och 2 gång prestationsmått, nämligen gångavstånd på 6 minuter samt metabola energiåtgång. Med hjälp av förklarbar AI påvisades det stora påverkan av muskelstyrka på både gångsträckan och energiåtgång. Gånghjälpmedlet samt kardiovaskulär kapaciteten hade också betydande påverkan på gångprestation. Enskilda analyser gav insiktsfull information om varje individ. Resultaten från dessa studier visar på potentiella tillämpningar av avancerad maskininlärning och AI metoder för att analysera komplexa relationer mellan funktion och motorisk prestation efter ryggmärgsskada. / <p>QC 20240221</p>