The standard camera- and force plate-based set-up for motion analysis suffers from the disadvantage of being limited to laboratory settings. Since adaptive algorithms are able to learn the connection between known inputs and outputs and generalise this knowledge to unknown data, these algorithms can be used to leverage motion analysis outside the laboratory. In most biomechanical applications, feedforward neural networks are used, although these networks can only work on time normalised data, while recurrent neural networks can be used for real time applications. Therefore, this study compares the performance of these two kinds of neural networks on the prediction of ground reaction force and joint moments of the lower limbs during gait based on joint angles determined by optical motion capture as input data. The accuracy of both networks when generalising to new data was assessed using the normalised root-mean-squared error, the root-mean-squared error and the correlation coefficient as evaluation metrics. Both neural networks demonstrated a high performance and good capabilities to generalise to new data. The mean prediction accuracy over all parameters applying a feedforward network was higher (r = 0.963) than using a recurrent long short-term memory network (r = 0.935).

The purpose of this systematic review is to analyse the results of operative treatment for midportion Achilles tendinopathy and to provide evidence based recommendation for the indication of the individual published techniques. MEDLINE, Cochrane Database, ISI Web of Knowledge and Google databases (1945 till September 2014) were electronically searched. The quality of the included articles was evaluated using the Coleman Methodology Score. Success rates, patient satisfaction, and the complication rates were determined. Twenty studies met our inclusion criteria. A total of 801 tendons were treated in 714 patients with open or minimally invasive techniques. The mean success rate was 83.4 %. Complications were reported in 6.3 % of the cases. The articles on minimally invasive techniques and open procedures reported on an average success rate of 83.6 % and 78.9 (p = 0.987). Patient satisfaction rates for minimally invasive techniques and open procedures were 78.5 % and 78.1 % (p = 0.211). The complication rate was 5.3 % for the minimally invasive techniques and 10.5 % for the open procedures (p = 0.053). We conclude that success rates of minimally invasive and open treatments are not different and that there is no difference in patient satisfaction but there is a tendency for more complications to occur in open procedures.

Background: The etiology of Anterior Cruciate Ligament (ACL) injury in women football results from the interaction of several extrinsic and intrinsic risk factors. Extrinsic factors change dynamically, also due to fatigue. However, existing biomechanical findings concerning the impact of fatigue on the risk of ACL injuries remains inconsistent. We hypothesized that fatigue induced by acute workload in short and intense game periods, might in either of two ways: by pushing lower limbs mechanics toward a pattern close to injury mechanism, or alternatively by inducing opposed protective compensatory adjustments.Aim: In this study, we aimed at assessing the extent to which fatigue impact on joints kinematics and kinetics while performing repeated changes of direction (CoDs) in the light of the ACL risk factors.Methods: This was an observational, cross-sectional associative study. Twenty female players (age: 20–31 years, 1st–2nd Italian division) performed a continuous shuttle run test (5-m) involving repeated 180°-CoDs until exhaustion. During the whole test, 3D kinematics and ground reaction forces were used to compute lower limb joints angles and internal moments. Measures of exercise internal load were: peak post-exercise blood lactate concentration, heart rate (HR) and perceived exertion. Continuous linear correlations between kinematics/kinetics waveforms (during the ground contact phase of the pivoting limb) and the number of consecutive CoD were computed during the exercise using a Statistical Parametric Mapping (SPM) approach.Results: The test lasted 153 ± 72 s, with a rate of 14 ± 2 CoDs/min. Participants reached 95% of maximum HR and a peak lactate concentration of 11.2 ± 2.8 mmol/L. Exercise duration was inversely related to lactate concentration (r = −0.517, p < 0.01), while neither%HR_max nor [La^–]_b nor RPE were correlated with test duration before exhaustion (p > 0.05). Alterations in lower limb kinematics were found in 100%, and in lower limb kinetics in 85% of the players. The most common kinematic pattern was a concurrent progressive reduction in hip and knee flexion angle at initial contact (10 players); 5 of them also showed a significantly more adducted hip. Knee extension moment decreased in 8, knee valgus moment increased in 5 players. A subset of participants showed a drift of pivoting limb kinematics that matches the known ACL injury mechanism; other players displayed less definite or even opposed behaviors.Discussion: Players exhibited different strategies to cope with repeated CoDs, ranging from protective to potentially dangerous behaviors. While the latter was not a univocal effect, it reinforces the importance of individual biomechanical assessment when coping with fatigue.

The use of machine learning to estimate joint angles from inertial sensors is a promising approach to in-field motion analysis. In this context, the simplification of the measurements by using a small number of sensors is of great interest. Neural networks have the opportunity to estimate joint angles from a sparse dataset, which enables the reduction of sensors necessary for the determination of all three-dimensional lower limb joint angles. Additionally, the dimensions of the problem can be simplified using principal component analysis. Training a long short-term memory neural network on the prediction of 3D lower limb joint angles based on inertial data showed that three sensors placed on the pelvis and both shanks are sufficient. The application of principal component analysis to the data of five sensors did not reveal improved results. The use of longer motion sequences compared to time-normalised gait cycles seems to be advantageous for the prediction accuracy, which bridges the gap to real-time applications of long short-term memory neural networks in the future.