Convolutional LSTM: a deep learning approach to predict shoulder joint reaction forces

Impact of specialized fatigue and backhand smash on the ankle biomechanics of female badminton players

During fatigued conditions, badminton players may experience adverse effects on their ankle joints during smash landings. In addition, the risk of ankle injury may vary with different landing strategies. This study aimed to investigate the influence of sport-specific fatigue factors and two backhand smash actions on ankle biomechanical indices. Thirteen female badminton players (age: 21.2 ± 1.9 years; height: 167.1 ± 4.1 cm; weight: 57.3 ± 5.1 kg; BMI: 20.54 ± 1.57 kg/m2) participated in this study. An 8-camera Vicon motion capture system and three Kistler force platforms were used to collect kinematic and kinetic data before and after fatigue for backhand rear-court jump smash (BRJS) and backhand lateral jump smash (BLJS). A 2 × 2 repeated measures analysis of variance was employed to analyze the effects of these smash landing actions and fatigue factors on ankle biomechanical parameters. Fatigue significantly affected the ankle-joint plantarflexion and inversion angles at the initial contact (IC) phase (p < 0.05), with both angles increasing substantially post-fatigue. From a kinetic perspective, fatigue considerably influenced the peak plantarflexion and peak inversion moments at the ankle joint, which resulted in a decrease the former and an increase in the latter after fatigue. The two smash landing actions demonstrated different landing strategies, and significant main effects were observed on the ankle plantarflexion angle, inversion angle, peak dorsiflexion/plantarflexion moment, peak inversion/eversion moment, and peak internal rotation moment (p < 0.05). The BLJS landing had a much greater landing inversion angle, peak inversion moment, and peak internal rotation moment compared with BRJS landing. The interaction effects of fatigue and smash actions significantly affected the muscle force of the peroneus longus (PL), with a more pronounced decrease in the force of the PL muscle post-fatigue in the BLJS action(post-hoc < 0.05). This study demonstrated that fatigue and smash actions, specifically BRJS and BLJS, significantly affect ankle biomechanical parameters. After fatigue, both actions showed a notable increase in IC plantarflexion and inversion angles and peak inversion moments, which may elevate the risk of lateral ankle sprains. Compared with BRJS, BLJS poses a higher risk of lateral ankle sprains after fatigue.

Glenohumeral joint force prediction with deep learning

Deep learning models (DLM) are efficient replacements for computationally intensive optimization techniques. Musculoskeletal models (MSM) typically involve resource-intensive optimization processes for determining joint and muscle forces. Consequently, DLM could predict MSM results and reduce computational costs. Within the total shoulder arthroplasty (TSA) domain, the glenohumeral joint force represents a critical MSM outcome as it can influence joint function, joint stability, and implant durability. Here, we aimed to employ deep learning techniques to predict both the magnitude and direction of the glenohumeral joint force. To achieve this, 959 virtual subjects were generated using the Markov-Chain Monte-Carlo method, providing patient-specific parameters from an existing clinical registry. A DLM was constructed to predict the glenohumeral joint force components within the scapula coordinate system for the generated subjects with a coefficient of determination of 0.97, 0.98, and 0.98 for the three components of the glenohumeral joint force. The corresponding mean absolute errors were 11.1, 12.2, and 15.0 N, which were about 2% of the maximum glenohumeral joint force. In conclusion, DLM maintains a comparable level of reliability in glenohumeral joint force estimation with MSM, while drastically reducing the computational costs.

Analysis of technical characteristics of typical lower limb balance movements in Tai Chi: a cross-sectional study based on AnyBody bone muscle modeling

BACKGROUND

Tai Chi is an excellent traditional Chinese physical fitness exercise, and its unique balancing movements are very important for improving human balance. In this study, the two most representative Tai Chi balance movements, "knee lift balance" and "leg stirrup balance", were selected to establish the lower limb bone muscle model of Tai Chi balance movements by using computer simulation modeling technology, aiming to analyze the characteristics of the lower limb movement mechanisms of typical balance movements, to provide a quantitative theoretical basis for improving the scientific level of Tai Chi.

METHOD

A total of 16 subjects were recruited for this study. the BTS three-dimensional motion capture system and three-dimensional force platform were used for motion data acquisition, the physiological electromyographic signals were collected using BTS surface electromyography, and the lower limb bone muscle model of Tai Chi balance movements was established by AnyBody human simulation.

RESULT

In the knee lift balancing movement, the balance leg hip abduction/adduction angle, hip flexion/extension moment, and the strength of the rectus femoris muscle, biceps femoris short capitis, and iliacus muscle of the amateur group was significantly smaller than that of the professional group (P < 0.01). In the leg stirrup balance movement, the knee flexion/extension angle of the balancing leg in the amateur group was significantly greater than that in the professional group (P < 0.01), and the hip flexion/extension angle, hip inversion/abduction angle, knee flexion/extension moment, hip flexion/extension moment, the strength iliacus, gluteus maximus, and obturator internus were significantly smaller than those in the professional group (P < 0.01). The integral EMG of the biceps femoris of the support leg in the amateur group was significantly smaller than that of the professional group (P < 0.01). The integral EMG of the lateral femoral muscle of the balance leg was significantly smaller than that of the professional group (P < 0.01).

CONCLUSION

In this study, we found that the balancing leg of the balancing movement has a larger hip joint angle, the stirrup balancing knee joint angle is smaller, and the hip and knee joint moments are larger. This is related to joint activity and muscle activation, and amateurs should pay attention to increasing the range of motion of the hip joint and decreasing the range of motion of the knee joint when practicing to better stimulate exercise of the lower limb joints. In addition, the practice of balancing movements should strengthen the iliacus muscle, which plays an important role in maintaining the stable balance of the lower limbs, and strengthen the knee flexor and extensor muscles and hip adductor/abductor muscles of the balancing leg, thus promoting the stability of the balancing leg movements.

Machine Learning for Optical Motion Capture-Driven Musculoskeletal Modelling from Inertial Motion Capture Data

Marker-based Optical Motion Capture (OMC) systems and associated musculoskeletal (MSK) modelling predictions offer non-invasively obtainable insights into muscle and joint loading at an in vivo level, aiding clinical decision-making. However, an OMC system is lab-based, expensive, and requires a line of sight. Inertial Motion Capture (IMC) techniques are widely-used alternatives, which are portable, user-friendly, and relatively low-cost, although with lesser accuracy. Irrespective of the choice of motion capture technique, one typically uses an MSK model to obtain the kinematic and kinetic outputs, which is a computationally expensive tool increasingly well approximated by machine learning (ML) methods. Here, an ML approach is presented that maps experimentally recorded IMC input data to the human upper-extremity MSK model outputs computed from ('gold standard') OMC input data. Essentially, this proof-of-concept study aims to predict higher-quality MSK outputs from the much easier-to-obtain IMC data. We use OMC and IMC data simultaneously collected for the same subjects to train different ML architectures that predict OMC-driven MSK outputs from IMC measurements. In particular, we employed various neural network (NN) architectures, such as Feed-Forward Neural Networks (FFNNs) and Recurrent Neural Networks (RNNs) (vanilla, Long Short-Term Memory, and Gated Recurrent Unit) and a comprehensive search for the best-fit model in the hyperparameters space in both subject-exposed (SE) as well as subject-naive (SN) settings. We observed a comparable performance for both FFNN and RNN models, which have a high degree of agreement (ravg,SE,FFNN=0.90±0.19, ravg,SE,RNN=0.89±0.17, ravg,SN,FFNN=0.84±0.23, and ravg,SN,RNN=0.78±0.23) with the desired OMC-driven MSK estimates for held-out test data. The findings demonstrate that mapping IMC inputs to OMC-driven MSK outputs using ML models could be instrumental in transitioning MSK modelling from 'lab to field'.