With a prevalence of 17% in men and 4% in women, Femoroacetabular Impingement (FAI) of type cam is characterized by a decreased femoral head-neck offset and/or asphericity of the lateral femoral head, associated with groin pain and reduced hip range of motion. Since the aetiology is still unclear, the mechanisms of development, progression and degeneration of FAI are largely investigated. Musculoskeletal modeling can support the development of a biomechanical framework to advance the research on FAI pathomechanisms, expand the knowledge about hip contact load distribution in FAI population, and relate the muscle and hip contact forces to the alterations observed during functional tasks. Therefore, this thesis is composed of two parts: the development of a methodological framework, and its application to the investigation of FAI pathomechanisms. The variability of the modelling outcomes (i.e.., body kinematics, torques, contact and muscle forces) to different marker sets, pelvic marker misplacements, and hip joint center (HJC) location was investigated within an inverse kinematic framework. The findings from such studies supported the modelling choices for the clinical investigation of FAI pathomechanisms. In particular, the performance of three different marker sets (Plug-in-Gait, University of Ottawa Motion Analysis Model and a 3-marker-cluster marker set) was compared, and absolute and relative reliability indices were calculated with the purpose of finding a simple yet reliable marker set to be used within an inverse kinematic framework in a clinical study. Thereafter, the sensitivity of joint angles, moments and hip contact forces to simulated inaccurate pelvic tilt was analyzed. The resulting variability indices were high with variations up to 1.3 times the body weight in hip contact forces. The kinematic variations propagated non-linearly to all planes and joints, showing the importance of adjusting possible pelvic misalignments. A methodology was presented to correct the pelvic alignment when the relative position of surface pelvic markers with respect to bony landmarks is known from medical images. The HJC location is a crucial modelling parameter in the analysis of hip kinematics and forces. A certain degree of customization could be introduced in the model by using HJC measured from medical images. Therefore, the performance of a generic musculoskeletal model with customized or non-customized HJC was compared during walking. Hip contact forces were highly sensitive to HJC location, especially because of the dependency of muscle moment arms to HJC changes. However, the variation of HJC without consistent muscle anatomy customization introduced artifacts that could potentially produce inaccurate muscle and joint contact forces estimation. When HJC cannot be measured from medical images, regression equations can be used instead. Therefore, the validity of two popular HJC regression equations (Harrington and Davis) was tested on FAI participants using non-parametric statistical and Bland-Altman tests. The results indicated that the equations were valid for FAI population. In addition, skin thickness measurements were provided for pelvic bony landmarks, and their correlation with body mass index was proposed for systematic error reduction. New adult-specific regression equations were developed from medical images. The described methodological framework was then applied to investigate the functional alterations observed in FAI population. The differences in muscle and hip contact forces were compared between FAI and healthy control groups during level walking. The FAI group showed reduced muscle and hip contact forces, which were linked to the lower normalized walking speed and shorter step length. These results can be interpreted as a protective mechanism developed by FAI patients to prevent high compression at the site of impingement, given that the compressing hip contact force was directed towards the anterior-superior quadrant of the acetabulum, consistent with the localization of the cam-type deformity and the cartilage and labrum damages. Based on these findings, a possible FAI pathomechanism was proposed, which could be used to support the development of preventive treatment and intervention for symptomatic FAI patients.

Differences between humans and extant apes in the pelvis and its key muscle attachment sites are thought to reflect a trade-off between arboreal and bipedal locomotor abilities. Human pelvic morphology enables the hamstrings to effectively power the hip hyperextension necessary for efficient bipedal locomotion, but this morphology is thought to reduce the capacity of these muscles to powerfully extend the hip when in a flexed position typical of arboreal locomotion. This research tested whether the enlarged human gluteus maximus may have been shaped by the continued importance of climbing among humans, as it has been suggested that it plays a compensatory role during powerful hip extension due to the reduced ability of the hamstrings. Musculoskeletal modeling and electromyography were used to assess the relative function of the gluteus maximus and the hamstrings in a human participant across two movement trials that required different amounts of hip extension: 1) bipedal walking, and 2) standing from a deep squat. It was hypothesized that the gluteus maximus would perform more effectively than the hamstrings to power hip extension from the flexed position of the squat. Differences in relative muscle activity across the two motions support this hypothesis in general, and implications for the evolutionary significance of the human gluteus maximus are that this muscle plays an important and likely compensatory role with the hamstrings during both standing up from a squat and bipedal walking. Results support the growing body of research that indicates that it is important to consider a broader range of human locomotive repertoires as of evolutionary significance, beyond solely terrestrial bipedal locomotion. / Graduate

[ES] El aumento de la esperanza de vida y la inversión de la pirámide poblacional plantean un desafío al sector sanitario debido al incremento de lesiones articulares y cirugías reconstructivas per cápita. Esto aumenta la demanda de personal rehabilitador en una sociedad con menos población activa. La robotización de las terapias puede aliviar esta presión al automatizar los ejercicios repetitivos. Además, el uso de tecnologías como los modelos músculo-esqueléticos mejoran las terapias, acortando los tiempos de recuperación y optimizan los resultados. La aplicación de modelos músculo-esqueléticos en robots de rehabilitación garantiza la seguridad del paciente al limitar las fuerzas excesivas y evitar posiciones peligrosas. Al mismo tiempo, brinda información adicional al personal rehabilitador para un seguimiento más preciso y personalizado de las terapias a los pacientes. Sin embargo, estos modelos son costosos computacionalmente, lo que dificulta su implementación en el control de robots. En el proyecto en el cual se integra esta tesis, se ha establecido el objetivo de construir un robot de rehabilitación que integre un modelo músculo-esquelético capaz de calcular en tiempo real las fuerzas musculares y articulares. Conocer las fuerzas garantiza la seguridad del paciente, proporciona información sobre las fuerzas durante los ejercicios y optimiza las terapias. Además, a partir del modelo músculo-esquelético se busca desarrollar nuevas herramientas para personalizar los ejercicios y mejorar los resultados. El modelo del miembro inferior, con seis grados de libertad, se ha simplificado para garantizar el cálculo en tiempo real. Se ha utilizado el concepto de grado de libertad funcional para predecir las relaciones entre los grados de libertad en un ejercicio concreto, reduciendo la carga computacional y permitiendo su uso en tiempo real durante los ejercicios de rehabilitación. El modelo consta de tres grados de libertad para la cadera, simulando una junta esférica, modelada como tres pares de revolución perpendiculares entre sí, uno para la rodilla, modelada como un mecanismo de cuatro barras, que simula el movimiento relativo entre el fémur y la tibia, y dos grados de libertad para el tobillo. Se calcula el centro de giro de la cadera y los parámetros del mecanismo de cuatro barras mediante ejercicios de calibración articular para lograr una mayor personalización del modelo. El tobillo se ha modelado empleando los datos de Klein Horsman debido a su dificultad de calibración. Del mismo trabajo se han tomado los parámetros musculares, los cuales se han simplificado para reducir el coste computacional. Se calculan las tensiones musculares mediante las condiciones de Karush-Kuhn-Tucker, minimizando el sumatorio cuadrático de las tensiones de los músculos. El modelo se ha validado y verificado siguiendo las recomendaciones de buenas prácticas de Hicks. Se han comparado los resultados del presente modelo con otro similar generado en AnyBody y con los datos empíricos del ``Grand Challenge", se ha analizado la solidez del modelo frente a las simplificaciones realizadas y los errores de los datos de entrada. Según los resultados obtenidos, el modelo músculo-esquelético es lo suficientemente preciso para ser utilizado en un robot de rehabilitación, garantizando la seguridad de los pacientes y prediciendo la activación muscular. Por último, se han desarrollado dos nuevas herramientas utilizando el modelo actual. La primera estima la Máxima Contracción Voluntaria del sujeto proyectando las fuerzas musculares al efector final del robot. La segunda herramienta calcula la fuerza externa necesaria para garantizar una fuerza muscular específica. Empleando ambas herramientas se logra una mayor personalización de las terapias de rehabilitación, mejorando el proceso. Ambas herramientas han sido probadas empleando el robot de rehabilitación. / [CA] L'augment de l'esperança de vida i la inversió de la piràmide poblacional plantegen un desafiament al sector sanitari a causa de l'increment de lesions articulars i cirurgies reconstructives. Això augmenta la demanda de personal rehabilitador en una societat amb menys població activa. La robotització de les teràpies pot alleujar aquesta pressió a l'autoritzar els exercisses repetitius. A més, l'ús de tecnologies com els models múscul-esquelètics milloren les teràpies, acurtant els temps de recuperació i optimitzant els resultats. L'aplicació de models múscul-esquelètics en robots de rehabilitació garanteix la seguretat del pacient en limitar les forces excessives i evitar posicions perilloses. Al mateix temps, ofereix informació addicional al personal rehabilitador per a un seguiment més precís i personalitzat de les teràpies als pacients. No obstant això, aquests models són costosos computacionalment, el que dificulta la seua implementació en el control de robots. En el projecte en el qual s'integra aquesta tesi, s'ha establit l'objectiu de construir un robot de rehabilitació que integre un model múscul-esquelètic capaç de calcular en temps real les forces musculars i articulars. Conéixer les forces garanteix la seguretat del pacient, proporcionant informació sobre les forces durant els exercicis i optimitza les teràpies. A més, a partir del model múscul-esquelètic es busca desenvolupar noves ferramentes per a personalitzar els exercicis millorar els resultats. El model del membre inferior, amb sis graus de llibertat, s'ha simplificat per garantir el càlcul en temps real. S'ha utilitzat el concepte de grau de llibertat funcional per a predir les relacions entre els graus de llibertat d'un exercici concret, reduint la càrrega computacional i permitent el seu ús en temps real durant els exercisses de rehabilitació. El model consta de tres graus de llibertat per al maluc, simulant una junta esfèrica, modelant com tres parells de revolucions perpendiculars entre si, un per al genoll, modelat com un mecanisme de quatre barres, què simula el moviment relatiu entre el fèmur i la tíbia, i dos graus de llibertat per al turmell. Es calcula el centre de gir del maluc i els paràmetres del mecanisme de quatre barres mitjançant exercicis de calibratge articular per a aconseguir una major personalització del model. El turmell s'ha modelat emprant les dades de Klein Horsman a causa de la seua dificultat de calibració. Del mateix treball s'han pres els paràmetres musculars, els quals s'han simplificat per a reduir el cost computacional. S'ha calculat les tensions musculars mitjançant les condicions de Karush-Kuhn-Tucker, minimitzant el sumatori quadràtic de les tensions musculars. El model s'ha validat i verificat seguint les recomanacions de bones pràctiques d'Hicks. S'ha comparat els resultats del present model amb altre similar generat en AnyBody i amb les dades empíriques del ''Grand Challenge", s'ha analitzat la solidesa del model enfront de les simplificacions realitzades i els errors de les dades d'entrada. Segons els resultats obtinguts, el model múscul-esquelètic és prou precís per a ser utilitzat en un robot de rehabilitació, garantint la seguretat dels pacients i predient l'activació muscular. En últim lloc, s'han desenvolupat dues noves ferramentes utilitzant el model actual. La primera estima la Màxima Contracció Voluntària del subjecte projectant les forces musculars a l'efector final del robot. La segona ferramenta calcula la força externa necessària per a garantir una força muscular específica. Emprant totes dues eines s'aconsegueix una major personalització de les teràpies de rehabilitació, millorant el procés. Totes dues ferramentes han sigut provades fent servir el robot de rehabilitació. / [EN] Increasing life expectancy and the inversion of the population pyramid pose a challenge to the healthcare sector due to the rise in joint injuries and reconstructive surgeries per capita. This increases the demand for rehabilitation personnel in a society with a smaller active population. The robotization of therapies can alleviate this pressure by automating repetitive exercises. Furthermore, the use of technologies like musculoskeletal models enhances therapies, reducing recovery times, and optimizing outcomes. The application of musculoskeletal models in rehabilitation robots ensures patient safety by limiting excessive forces and avoiding dangerous positions. At the same time, it provides additional information to rehabilitation personnel for more precise and personalized monitoring of patient therapies. However, these models have a high computational cost, which makes their implementation in robot control challenging. In the project that integrates this thesis, the objective has been set to build a rehabilitation robot that incorporates a musculoskeletal model capable of real-time calculation of muscular and joint forces. Understanding these forces ensures patient safety, provides information about forces during exercises, and optimizes therapies. Additionally, based on the musculoskeletal model, new tools are being developed to personalize exercises and improve results. The lower limb model, with six degrees of freedom, has been simplified to enable real-time calculation. The concept of functional degrees of freedom has been used to predict the relationships between degrees of freedom in a specific exercise, reducing the computational burden and enabling real-time use during rehabilitation exercises. The model consists of three degrees of freedom for the hip, simulating a spherical joint, modeled as three sets of perpendicular revolute pairs, one for the knee, modeled as a four-bar mechanism, that simulates the relative movement between the femur and the tibia, and two degrees of freedom for the ankle. The hip's pivot center and the parameters of the four-bar mechanism have been calculated through joint calibration exercises to achieve a higher level of model personalization. The ankle has been modeled using Klein Horsman's data due to its calibration complexity. Muscle parameters from the same work have been taken and simplified to reduce computational costs. Muscle tensions are calculated using the Karush-Kuhn-Tucker conditions, minimizing the squared sum of muscle tensions. The model has been validated and verified following Hicks' best practices recommendations. The results of this model have been compared with a similar one generated in AnyBody and with empirical data from the ``Grand Challenge". The model's robustness against the made simplifications and input data errors has been analyzed. According to the obtained results, the musculoskeletal model is accurate enough to be used in a rehabilitation robot, ensuring patient safety and predicting muscle activation. Finally, two new tools have been developed using the current model. The first one estimates the Maximum Voluntary Contraction of the subject by projecting muscular forces to the robot's end effector. The second tool calculates the external force required to achieve a specific muscular force. By employing both tools, greater personalization of rehabilitation therapies is achieved, improving the process. Both tools have been tested using the rehabilitation robot. / Esta tesis y la investigación realizada han sido financiadas por la Agencia Estatal de Investigación con los proyectos “Integración de modelos biomecánicos en el desarrollo y operación de robots rehabilitadores reconfigurables” (DPI2017-84201-R-AR) y “Sistema robótico paralelo con control basado en modelo músculo-esquelético para la monitorización y entrenamiento del sistema propioceptivo” (PID2021-125694OB-I00). El autor de la presente tesis recibió la beca de la Agencia Estatal de Investigación: Ayuda para contrato predoctoral para la formación de doctores Zamora Ortiz, Pau. (PRE2018-083847). Con la cual ha podido dedicarse a tiempo completo a la formación e investigación que ha dado como fruto esta tesis doctoral. / Zamora Ortiz, P. (2023). Modelo músculo-esquelético del miembro inferior para rehabilitación con robot paralelo [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/199483