Prediction of caspase cleavage sites using Bayesian bio-basis function neural networks

MOTIVATION: Apoptosis has drawn the attention of researchers because of its importance in treating some diseases through finding a proper way to block or slow down the apoptosis process. Having understood that caspase cleavage is the key to apoptosis, we find novel methods or algorithms are essential for studying the specificity of caspase cleavage activity and this helps the effective drug design. As bio-basis function neural networks have proven to outperform some conventional neural learning algorithms, there is a motivation, in this study, to investigate the application of bio-basis function neural networks for the prediction of caspase cleavage sites. RESULTS: Thirteen protein sequences with experimentally determined caspase cleavage sites were downloaded from NCBI. Bayesian bio-basis function neural networks are investigated and the comparisons with single-layer perceptrons, multilayer perceptrons, the original bio-basis function neural networks and support vector machines are given. The impact of the sliding window size used to generate sub-sequences for modelling on prediction accuracy is studied. The results show that the Bayesian bio-basis function neural network with two Gaussian distributions for model parameters (weights) performed the best and the highest prediction accuracy is 97.15 +/- 1.13%. AVAILABILITY: The package of Bayesian bio-basis function neural network can be obtained by request to the author.     

Prediction of ground reaction forces during gait based on kinematics and a neural network model

Kinetic information during human gait can be estimated with inverse dynamics, which is based on anthropometric, kinematic, and ground reaction data. While collecting ground reaction data with a force plate is useful, it is costly and requires regulated space. The goal of this study was to propose a new, accurate methodology for predicting ground reaction forces (GRFs) during level walking without the help of a force plate. To predict GRFs without a force plate, the traditional method of Newtonian mechanics was used for the single support phase. In addition, an artificial neural network (ANN) model was applied for the double support phase to solve statically indeterminate structure problems. The input variables of the ANN model, which were selected to have both dependency and independency, were limited to the trajectory, velocity, and acceleration of the whole segment's mass centre to minimise errors. The predicted GRFs were validated with actual GRFs through a ten-fold cross-validation method, and the correlation coefficients (R) for the ground forces were 0.918 in the medial–lateral axis, 0.985 in the anterior–posterior axis, and 0.991 in the vertical axis during gait. The ground moments were 0.987 in the sagittal plane, 0.841 in the frontal plane, and 0.868 in the transverse plane during gait. The high correlation coefficients(R) are due to the improvement of the prediction rate in the double support phase. This study also proved the possibility of calculating joint forces and moments based on the GRFs predicted with the proposed new hybrid method. Data generated with the proposed method may thus be used instead of raw GRF data in gait analysis and in calculating joint dynamic data using inverse dynamics.     

Estimating the complete ground reaction forces with pressure insoles in walking

This study presented a method to estimate the complete ground reaction forces from pressure insoles in walking. Five male subjects performed 10 walking trials in a laboratory. The complete ground reaction forces were collected during a right foot stride by a force plate at 1000Hz. Simultaneous plantar pressure data were collected at 100Hz by a pressure insole system with 99 sensors covering the whole plantar area. Stepwise linear regressions were performed to individually reconstruct the complete ground reaction forces in three directions from the 99 individual pressure data until redundancy among the predictors occurred. An additional linear regression was performed to reconstruct the vertical ground reaction force by the sum of the value of the 99 pressure sensors. Five other subjects performed the same walking test for validation. Estimated ground reaction forces in three directions were calculated with the developed regression models, and were compared with the real data recorded from force plate. Accuracy was represented by the correlation coefficient and the root mean square error. Results showed very good correlation in anterior-posterior (0.928) and vertical (0.989) directions, and reasonable correlation in medial-lateral direction (0.719). The root mean square error was about 12%, 5% and 28% of the peak recorded value. Future studies should aim to generalize the methods or to establish specific methods to other subjects, patients, motions, footwear and floor conditions. The method gives an extra option to study an estimation of the complete ground reaction forces in any environment without the constraints from the number and location of force plates.     

Effects of a knee extension constraint brace on selected lower extremity motion patterns during a stop-jump task

Small knee flexion angle during landing has been proposed as a potential risk factor for sustaining noncontact ACL injury. A brace that promotes increased knee flexion and decreased posterior ground reaction force during landing may prove to be advantageous for developing prevention strategies. Forty male and forty female recreational athletes were recruited. Three-dimensional videographic and ground reaction force data in a stop-jump task were collected in three conditions. Knee flexion angle at peak posterior ground reaction force, peak posterior ground reaction force, the horizontal velocity of approach run, the vertical velocity at takeoff, and the knee flexion angle at takeoff were compared among conditions: knee extension constraint brace, nonconstraint brace, and no brace. The knee extension constraint brace significantly increased knee flexion angle at peak posterior ground reaction force. Both knee extension constraint brace and nonconstraint brace significantly decreased peak posterior ground reaction force during landing. The brace and knee extension constraint did not significantly affect the horizontal velocity of approach run, the vertical velocity at takeoff, and the knee flexion angle at takeoff. A knee extension constraint brace exhibits the ability to modify the knee flexion angle at peak posterior ground reaction force and peak posterior ground reaction force during landing.     

Estimating the muscle forces generated in the human lower extremity when walking: a physiological solution

A model capable of estimating individual muscle activity during human lower extremity activity has been developed. This model was implemented and applied to normal gait. Both 3D cinematography and force platform records were collected to define the kinematics of the lower limb segments and the ground reaction forces. From these data the lengths, velocities, and moment arms of the muscles and the net muscle moments were calculated. These data were then used to estimate the output from a set of pattern generators which specified the neural input to and hence the force generated in each muscle. Results were obtained which demonstrate that it is feasible to construct a hierarchical physiological model which will estimate individual muscle forces. This statement must be tempered somewhat in that better anatomical and neurophysiological data must be made available and that this model, and all others attempting to estimate individual muscle forces, must be rigorously tested before being applied in a research, sports, or clinical setting.     

Comparison of in vivo measured loads in knee,hip and spinal implants during level walking

Walking is a task that we seek to understand because it is the most relevant human locomotion. Walking causes complex loading patterns and high load magnitudes within the human body. This work summarizes partially published load data collected in earlier in vivo measurement studies on 9 patients with telemeterized knee endoprostheses, 10 with hip endoprostheses and 5 with vertebral body replacements. Moreover, for the 19 endoprosthesis patients, additional simultaneously measured and previously unreported ground reaction forces are presented.The ground reaction force and the implant forces in the knee and hip exhibited a double peak during each step. The maxima of the ground reaction forces ranged from 100% to 126% bodyweight. In comparison, the greatest implant forces in the hip (249% bodyweight) and knee (271% bodyweight) were much greater. The mean peak force measured in the vertebral body replacement was 39% bodyweight and occurred at different time points of the stance phase.We concluded that walking leads to high load magnitudes in the knee and hip, whereas the forces in the vertebral body replacement remained relatively low. This indicates that the first peak force was greater in the hip than in the knee joint while this was reversed for the second peak force. The forces in the spinal implant were considerably lower than in the knee and hip joints.     

Quantifying plyometric intensity via rate of force development, knee joint, and ground reaction forces

Because the intensity of plyometric exercises usually is based simply upon anecdotal recommendations rather than empirical evidence, this study sought to quantify a variety of these exercises based on forces placed upon the knee. Six National Collegiate Athletic Association Division I athletes who routinely trained with plyometric exercises performed depth jumps from 46 and 61 cm, a pike jump, tuck jump, single-leg jump, countermovement jump, squat jump, and a squat jump holding dumbbells equal to 30% of 1 repetition maximum (RM). Ground reaction forces obtained via an AMTI force plate and video analysis of markers placed on the left hip, knee, lateral malleolus, and fifth metatarsal were used to estimate rate of eccentric force development (E-RFD), peak ground reaction forces (GRF), ground reaction forces relative to body weight (GRF/BW), knee joint reaction forces (K-JRF), and knee joint reaction forces relative to body weight (K-JRF/BW) for each plyometric exercise. One-way repeated measures analysis of variance indicated that E-RFD, K-JRF, and K-JRF/BW were different across the conditions (p < 0.05), but peak GRF and GRF/BW were not (p > 0.05). Results indicate that there are quantitative differences between plyometric exercises in the rate of force development during landing and the forces placed on the knee, though peak GRF forces associated with landing may not differ.     

Gene network inference using continuous time Bayesian networks: a comparative study and application to Th17 cell differentiation

Background

Dynamic aspects of gene regulatory networks are typically investigated by measuring system variables at multiple time points. Current state-of-the-art computational approaches for reconstructing gene networks directly build on such data, making a strong assumption that the system evolves in a synchronous fashion at fixed points in time. However, nowadays omics data are being generated with increasing time course granularity. Thus, modellers now have the possibility to represent the system as evolving in continuous time and to improve the models’ expressiveness.

Results

Continuous time Bayesian networks are proposed as a new approach for gene network reconstruction from time course expression data. Their performance was compared to two state-of-the-art methods: dynamic Bayesian networks and Granger causality analysis. On simulated data, the methods comparison was carried out for networks of increasing size, for measurements taken at different time granularity densities and for measurements unevenly spaced over time. Continuous time Bayesian networks outperformed the other methods in terms of the accuracy of regulatory interactions learnt from data for all network sizes. Furthermore, their performance degraded smoothly as the size of the network increased. Continuous time Bayesian networks were significantly better than dynamic Bayesian networks for all time granularities tested and better than Granger causality for dense time series. Both continuous time Bayesian networks and Granger causality performed robustly for unevenly spaced time series, with no significant loss of performance compared to the evenly spaced case, while the same did not hold true for dynamic Bayesian networks. The comparison included the IRMA experimental datasets which confirmed the effectiveness of the proposed method. Continuous time Bayesian networks were then applied to elucidate the regulatory mechanisms controlling murine T helper 17 (Th17) cell differentiation and were found to be effective in discovering well-known regulatory mechanisms, as well as new plausible biological insights.

Conclusions

Continuous time Bayesian networks were effective on networks of both small and large size and were particularly feasible when the measurements were not evenly distributed over time. Reconstruction of the murine Th17 cell differentiation network using continuous time Bayesian networks revealed several autocrine loops, suggesting that Th17 cells may be auto regulating their own differentiation process.     

Kinetic analysis of ski turns based on measured ground reaction forces

The objective of this study was to devise a method of kinetic analysis of the ground reaction force that enables the durations and magnitudes of forces acting during the individual phases of ski turns to be described exactly. The method is based on a theoretical analysis of physical forces acting during the ski turn. Two elementary phases were defined: (1) preparing to turn (initiation) and (2) actual turning, during which the center of gravity of the skier-ski system moves along a curvilinear trajectory (steering). The starting point of the turn analysis is a dynamometric record of the resultant acting ground reaction force applied perpendicularly on the ski surface. The method was applied to six expert skiers. They completed a slalom course comprising five gates arranged on the fall line of a 26° slope at a competition speed using symmetrical carving turns (30 evaluated turns). A dynamometric measurement system was placed on the carving skis (168 cm long, radius 16 m, data were recorded at 100 Hz). MATLAB procedures were used to evaluate eight variables during each turn: five time variables and three force variables. Comparison of the turn analysis results between individuals showed that the method is useful for answering various research questions associated with ski turns.     

Comparing jumping ability among athletes of various sports: vertical drop jumping from 60 centimeters

Drop jumping performance (DJP) is of high importance in order to achieve sporting performance in both team and individual sports. The purpose of the present study was to compare DJP among athletes from various sports. One hundred thirty-eight male athletes (age: 22.3 +/- 3.6 years, body height: 1.87 +/- 0.08 m, body mass: 81.8 +/- 10.8 kg) from 6 different sports performed drop jumps from 60 cm (DJ60) on a force plate. Results revealed that volleyball players jumped higher (p < 0.001) than other athletes. However, track and field athletes produced higher peak force and higher power output using a shorter upward phase (p < 0.001). Further examination using principal components analysis (PCA) revealed that team sport athletes and single scull rowers exhibited DJP utilizing force and time parameters differently than track and field athletes. Conclusively, DJP was different among athletes of various sports. Furthermore, PCA can be a useful method for evaluating the above mentioned differences and for monitoring drop jumping training programs.     

Implementing Gaussian process inference with neural networks

Gaussian processes compare favourably with backpropagation neural networks as a tool for regression, and Bayesian neural networks have Gaussian process behaviour when the number of hidden neurons tends to infinity. We describe a simple recurrent neural network with connection weights trained by one-shot Hebbian learning. This network amounts to a dynamical system which relaxes to a stable state in which it generates predictions identical to those of Gaussian process regression. In effect an infinite number of hidden units in a feed-forward architecture can be replaced by a merely finite number, together with recurrent connections.     

The effect of a repetitive, fatiguing lifting task on horizontal ground reaction forces

There are many outdoor work environments that involve the combination of repetitive, fatiguing lifting tasks and less-than-optimal footing (muddy/slippery ground surfaces). The focus of the current research was to evaluate the effects of lifting-induced fatigue of the low back extensors on lifting kinematics and ground reaction forces. Ten participants performed a repetitive lifting task over a period of 8 minutes. As they performed this task, the ground reaction forces and whole body kinematics were captured using a force platform and magnetic motion tracking system, respectively. Fatigue was verified in this experiment by documenting a decrease in the median frequency of the bilateral erector spinae muscles (pretest-posttest). Results indicate significant (p < 0.05) increases in the magnitude of the peak anterior/posterior (increased by an average of 18.3%) and peak lateral shear forces (increased by an average of 24.3%) with increasing time into the lifting bout. These results have implications for work environments such as agriculture and construction, where poor footing conditions and requirements for considerable manual materials handling may interact to create an occupational scenario with an exceptionally high risk of a slip and fall.     

A probabilistic method to estimate gait kinetics in the absence of ground reaction force measurements

Human joint torques during gait are usually computed using inverse dynamics. This method requires a skeletal model, kinematics and measured ground reaction forces and moments (GRFM). Measuring GRFM is however only possible in a controlled environment. This paper introduces a probabilistic method based on probabilistic principal component analysis to estimate the joint torques for healthy gait without measured GRFM. A gait dataset of 23 subjects was obtained containing kinematics, measured GRFM and joint torques from inverse dynamics in order to obtain a probabilistic model. This model was then used to estimate the joint torques of other subjects without measured GRFM. Only kinematics, a skeletal model and timing of gait events are needed. Estimation only takes 0.28 ms per time instant. Using cross-validation, the resulting root mean square estimation errors for the lower-limb joint torques are found to be approximately 0.1 Nm/kg, which is 6–18% of the range of the ground truth joint torques. Estimated joint torque and GRFM errors are up to two times smaller than model-based state-of-the-art methods. Model-free artificial neural networks can achieve lower errors than our method, but are less repeatable, do not contain uncertainty information on the estimates and are difficult to use in situations which are not in the learning set. In contrast, our method performs well in a new situation where the walking speed is higher than in the learning dataset. The method can for example be used to estimate the kinetics during overground walking without force plates, during treadmill walking without (separate) force plates and during ambulatory measurements.     

Pattern of anterior cruciate ligament force in normal walking

The goal of this study was to calculate and explain the pattern of anterior cruciate ligament (ACL) loading during normal level walking. Knee-ligament forces were obtained by a two-step procedure. First, a three-dimensional (3D) model of the whole body was used together with dynamic optimization theory to calculate body-segmental motions, ground reaction forces, and leg-muscle forces for one cycle of gait. Joint angles, ground reaction forces, and muscle forces obtained from the gait simulation were then input into a musculoskeletal model of the lower limb that incorporated a 3D model of the knee. The relative positions of the femur, tibia, and patella and the forces induced in the knee ligaments were found by solving a static equilibrium problem at each instant during the simulated gait cycle. The model simulation predicted that the ACL bears load throughout stance. Peak force in the ACL (303 N) occurred at the beginning of single-leg stance (i.e., contralateral toe off). The pattern of ACL force was explained by the shear forces acting at the knee. The balance of muscle forces, ground reaction forces, and joint contact forces applied to the leg determined the magnitude and direction of the total shear force acting at the knee. The ACL was loaded whenever the total shear force pointed anteriorly. In early stance, the anterior shear force from the patellar tendon dominated the total shear force applied to the leg, and so maximum force was transmitted to the ACL at this time. ACL force was small in late stance because the anterior shear forces supplied by the patellar tendon, gastrocnemius, and tibiofemoral contact were nearly balanced by the posterior component of the ground reaction.     

Using force sensing insoles to predict kinetic knee symmetry during a stop jump

Knee kinetic asymmetries are present during jump-landings in athletes returning to sport following anterior cruciate ligament (ACL) reconstruction, and are associated with an increased risk for sustaining a second ACL injury. The loadsol® is a wireless load sensing insole that can be used in non-laboratory settings. The purpose of this study was to determine if the loadsol® could be used to predict knee extension moment and power symmetry during a bilateral stop jump task in healthy recreational athletes. Forty-two uninjured recreational athletes completed seven bilateral stop jumps. During each landing, the loadsol® (100 Hz) measured plantar load while 3D ground reaction forces (1920 Hz) and lower extremity kinematics (240 Hz) were collected simultaneously. Peak impact force, loading rate, and impulse were quantified using the loadsol® and peak knee extension moment, average knee extension moment, and total knee work was quantified using the laboratory instrumentation. Limb symmetry indices were quantified for each outcome measure. Multivariate backwards regressions were used to determine if loadsol® symmetry could predict knee kinetic symmetry. Intraclass correlation coefficients (ICCs) and Bland-Altman plots were used to determine the agreement and error between predicted and actual knee kinetic symmetry. Loadsol® impulse and peak impact force symmetry significantly predicted kinetic knee symmetry and explained 42–61% of its variance. There was good agreement (ICCs = 0.742–0.862) between predicted and actual knee kinetic symmetry, and the error in the predicted outcomes range from ±18 to ±43. These results support using the loadsol® to screen for kinetic symmetries during landing in athletes following ACL reconstruction.     

A treadmill-mounted force platform

Muscle, bone, and tendon forces; the movement of the center of mass, and the spring properties of the body during terrestrial locomotion can be measured using ground-mounted force platforms. These measurements have been extremely time consuming because of the difficulty in obtaining repeatable constant speed trials (particularly with animals). We have overcome this difficulty by mounting a force platform directly under the belt of a motorized treadmill. With this arrangement, vertical force can be recorded from an unlimited number of successive ground contacts in a much shorter time. With this treadmill-mounted force platform it is possible to accurately make the following measurements over the full range of steady speeds and under various perturbations of normal gait: 1) vertical ground reaction force over the course of the contact phase; 2) peak forces in bone, muscle, and tendon; 3) the vertical displacement of the center of mass; and 4) contact time for the limbs. In our treadmill-force platform design, belt forces and frictional forces cause no measurable cross-talk problem. Natural frequency (160 Hz), nonlinearity (less than 5%), and position independence (less than 2%) are all quite acceptable. Motor-caused vibrations are greater than 150 Hz and thus can be easily filtered.     

Relationship of jumping and agility performance in female volleyball athletes

Court sports often require more frequent changes of direction (COD) than field sports. Most court sports require 180 degrees turns over a small distance, so COD in such sports might be best evaluated with an agility test involving short sprints and sharp turns. The purposes of this study were to (a) quantify vertical and horizontal force during a COD task, (b) identify possible predictors of court-sport-specific agility performance, and (c) examine performance difference between National Collegiate Athletic Association Division I, II, and III athletes. Twenty-nine collegiate female volleyball players completed a novel agility test, countermovement (CM) and drop jump tests, and an isometric leg extensor test. The number of athletes by division was as follows: I (n = 9), II (n = 11), and III (n = 9). The agility test consisted of 4 5-meter sprints with 3 180 degrees turns, including 1 on a multiaxial force platform so that the kinetic properties of the COD could be identified. One-way analysis of variance revealed that Division I athletes had significantly greater countermovement jump heights than Division III, and the effect size comparisons (Cohen's d) showed large-magnitude differences between Division I and both Divisions II and III for jump height. No other differences in performance variables were noted between divisions, although effect sizes reached moderate values for some comparisons. Regression analysis revealed that CM displacement was a significant predictor of agility performance, explaining approximately 34% of the variance. Vertical force was found to account for much of the total force exerted during the contact phase of the COD task, suggesting that performance in the vertical domain may limit the COD task used herein. This study indicates that individuals with greater CM performance also have quicker agility times and suggests that training predominantly in the vertical domain may also yield improvements in certain types of agility performance. This may hold true even if such agility performance requires a horizontal component.     

A new training algorithm using artificial neural networks to classify gender-specific dynamic gait patterns

The aim of this study was to present a new training algorithm using artificial neural networks called multi-objective least absolute shrinkage and selection operator (MOBJ-LASSO) applied to the classification of dynamic gait patterns. The movement pattern is identified by 20 characteristics from the three components of the ground reaction force which are used as input information for the neural networks in gender-specific gait classification. The classification performance between MOBJ-LASSO (97.4%) and multi-objective algorithm (MOBJ) (97.1%) is similar, but the MOBJ-LASSO algorithm achieved more improved results than the MOBJ because it is able to eliminate the inputs and automatically select the parameters of the neural network. Thus, it is an effective tool for data mining using neural networks. From 20 inputs used for training, MOBJ-LASSO selected the first and second peaks of the vertical force and the force peak in the antero-posterior direction as the variables that classify the gait patterns of the different genders.     

Ground reaction forces during the power clean

Identifying and understanding the key biomechanical factors that exemplify the power clean can provide athletes the proper tools needed to prevail at a competitive event. Therefore, the purpose of this study was to characterize and describe ground reaction forces (Fz) during the power clean lift. Three 60-Hz motion-detecting cameras and an AMTI force plate were used to collect data from 10 collegiate weightlifting men who performed a power clean at 60 and 70% of their last competitive maximum clean. The results revealed that a greater peak force (Fz) was produced during the second pull compared with the first pull and unweighted phases in both percentage lifts. As the system weight increased from 60 to 70%, the peak force (Fz) increased for the first pull and unweighted phases and decreased during the second pull phase. Learning the proper technique of the power clean may provide athletes the basic understanding needed to be competitive in a weightlifting or sporting event.     

Lower extremity joint torque predicted by using artificial neural network during vertical jump

The purpose of this study was to develop an artificial neural network (ANN) for predicting lower extremity joint torques using the ground reaction force (GRF) and related parameters derived by the GRF during counter-movement jump (CMJ) and squat jump (SJ). Ten student athletes performed CMJ and SJ. Force plate and kinematic data were recorded. Joint torques were calculated using inverse dynamics and ANN. We used a fully connected, feed-forward network. The network comprised of one input layer, one hidden layer and one output layer. It was trained by error back-propagation algorithm using Steepest Descent Method. Input parameters of the ANN were GRF measurements and related parameters. Output parameters were three lower extremity joint torques. ANN model fitted well with the results of the inverse dynamics output. Our observations indicate that the model developed in this study can be used to estimate three lower extremity joint torques for CMJ and SJ based on ground reaction force data and related parameters.