Calibrated EMG-Informed Neuromusculoskeletal Modelling to Estimate Physiologically Plausible Hip Joint Contact Forces in People with Hip Osteoarthritis

Abstract

Osteoarthritis (OA) is a common and progressive joint disease with a prevalence of 3.6-4.1% of the global population in 2010 and one of the leading causes of worldwide disability. Hip OA is less common than knee OA; although the prevalence of hip OA is 8% in people aged 85 years and over. There is no cure for hip OA, and pain and disability can be only managed through conservative therapies, or finally total hip replacement. Inappropriate hip joint loading, measured as hip joint contact forces (HJCF), during daily activities is believed to be a factor in hip OA initiation and progression. Neuromusculoskeletal (NMS) models, which are anatomical and physiological mathematical representations of an individual, can be used to estimate muscle-tendon and internal joint contact forces (JCF) during human movement. However, the musculoskeletal system is inherently indeterminant and there are infinite combinations of muscle-tendon forces that can produce the same external joint loads. Two main neural control solutions that have been employed to solve this muscle-tendon force distribution problem: optimisation and electromyography (EMG)-informed solutions. Although static optimisation is commonly used, an EMG-informed NMS approach may offer more physiologically plausible HJCF estimates, since it accounts for an individual’s unique muscle activation patterns. Indeed, individuals with hip OA have abnormal muscle activation, which may influence the HJCF. Calibration, or tuning, of the neuromuscular parameters in EMG-informed NMS modelling is required to create subject-specific models. There are various cost functions used for calibration and their selection is crucial for obtaining physiological plausible estimates of JCF. EMG-informed NMS modelling is driven by EMG signals experimentally measured with surface electrodes. However, surface electrodes cannot record deep muscles important for hip function, which limits this application of EMG-informed modelling. Recently, EMG-hybrid and EMG-assisted modes have been developed to address this limitation, although, to date, there have been only two limited studies using these EMG-informed approaches for the hip, and only in healthy individuals. Furthermore, no studies have assessed how different calibrations cost functions and EMG-informed neural control solutions affect HJCF estimates. Subsequently, this thesis aimed to assess the use of calibrated EMG-informed NMS modelling, available in the Calibrated EMG-informed Neuromusculoskeletal Modelling toolbox (CEINMS), to explore the HJCF from healthy individuals and patients with mild-to-moderate hip OA during walking. The first study compared how different calibration cost functions and EMG-informed neural solution modes affected the estimated HJCF from a healthy population. Calibrating with a cost function to minimise joint moments prediction errors and peak HJCF, used together with the EMG-assisted neural solution mode well tracked the external joint moments and measured EMGs, and was the recommended approach to calculate physiologically plausible HJCF. The study has been submitted to the Journal of Biomechanics with following author order and title: Hoang, H.X., Pizzolato, C., Diamond, L.E., Lloyd, D.G., 2017. “Subject-specific calibration of neuromuscular parameters enables neuromusculoskeletal models to estimate physiologically plausible hip joint contact forces in healthy adults.” The second study compared estimated HJCF in people with mild-to-moderate hip OA obtained with the calibrated EMG-assisted pipeline in CEINMS (recommendation from the first study) and the static optimisation pipeline in OpenSim. The results showed that EMG-assisted mode and static optimisation both well tracked lower limb joint moments. However, the EMG-assisted mode was able to track hip muscle co-contraction patterns, whereas static optimisation did not. The associated manuscript is in preparation for submission to the Journal of Biomechanics as: Hoang, H.X., Diamond, L.E., Lloyd, D.G., Pizzolato, C., 2017. “A calibrated EMG-informed neuromusculoskeletal modelling method can appropriately account for muscle co-contraction to estimate hip joint contact forces in people with hip osteoarthritis.” The third study assessed the hip muscle co-contraction and HJCF in people with mild-to-moderate hip OA and healthy individuals using the calibrated EMG-informed NMS modelling approach as recommended in the previous two studies. Higher hip muscle co-contraction and lower HJCF were observed in individuals with hip OA compared to controls. These findings challenge previous beliefs of over-loading due to muscle co-contraction in people with hip OA. The associated manuscript is in preparation for submission to Osteoarthritis and Cartilage as: Hoang H.X., Loureiro, A., Constantinou, M.,Barrett, R., Pizzolato C., Lloyd D.G., Diamond L.E., 2017. “People with symptomatic mild-to-moderate hip osteoarthritis exhibit higher muscle co-contraction and walk with lower hip joint contact forces compared to healthy people.” The findings of this thesis showed that (i) the neural solution control is essential when estimating muscle-tendon forces and HJCF with NMS models and, therefore, subject-specific muscle activation patterns should be included in any modelling framework, (ii) calibration through tracking of joint moments and peak HJCF in conjunction with the EMG-assisted neural solution mode resulted in more physiologically plausible HJCF compared to other methods, and (iii) people with mild-to-moderate hip OA walk with more hip muscle co-contraction and lower HJCF compared to healthy people.