A Deep Learning Modeling Framework to Capture Mixing Patterns in Reactive-Transport Systems

Prediction and control of chemical mixing are vital for many scientific areas such as subsurface reactive transport, climate modeling, combustion, epidemiology, and pharmacology. Due to the complex nature of mixing in heterogeneous and anisotropic media, the mathematical models related to this phenomenon are not analytically tractable. Numerical simulations often provide a viable route to predict chemical mixing accurately. However, contemporary modeling approaches for mixing cannot utilize available spatial-temporal data to improve the accuracy of the future prediction and can be compute-intensive, especially when the spatial domain is large and for long-term temporal predictions. To address this knowledge gap, we will present in this paper a deep learning (DL) modeling framework applied to predict the progress of chemical mixing under fast bimolecular reactions. This framework uses convolutional neural networks (CNN) for capturing spatial patterns and long short-term memory (LSTM) networks for forecasting temporal variations in mixing. By careful design of the framework—placement of non-negative constraint on the weights of the CNN and the selection of activation function, the framework ensures non-negativity of the chemical species at all spatial points and for all times. Our DL-based framework is fast, accurate, and requires minimal data for training. The time needed to obtain a forecast using the model is a fraction ($≈ \mathcal{O}(10^{−6}))$ of the time needed to obtain the result using a high-fidelity simulation. To achieve an error of 10% (measured using the infinity norm) for capturing local-scale mixing features such as interfacial mixing, only 24% to 32% of the sequence data for model training is required. To achieve the same level of accuracy for capturing global-scale mixing features, the sequence data required for model training is 64% to 70% of the total spatial-temporal data. Hence, the proposed approach—a fast and accurate way to forecast long-time spatial-temporal mixing patterns in heterogeneous and anisotropic media—will be a valuable tool for modeling reactive-transport in a wide range of applications.     

Spatio-temporal structure of cell distribution in cortical bone multicellular units: a mathematical model

Bone remodelling maintains the functionality of skeletal tissue by locally coordinating bone-resorbing cells (osteoclasts) and bone-forming cells (osteoblasts) in the form of Bone Multicellular Units (BMUs). Understanding the emergence of such structured units out of the complex network of biochemical interactions between bone cells is essential to extend our fundamental knowledge of normal bone physiology and its disorders. To this end, we propose a spatio-temporal continuum model that integrates some of the most important interaction pathways currently known to exist between cells of the osteoblastic and osteoclastic lineage. This mathematical model allows us to test the significance and completeness of these pathways based on their ability to reproduce the spatio-temporal dynamics of individual BMUs. We show that under suitable conditions, the experimentally observed structured cell distribution of cortical BMUs is retrieved. The proposed model admits travelling-wave-like solutions for the cell densities with tightly organised profiles, corresponding to the progression of a single remodelling BMU. The shapes of these spatial profiles within the travelling structure can be linked to the intrinsic parameters of the model such as differentiation and apoptosis rates for bone cells. In addition to the cell distribution, the spatial distribution of regulatory factors can also be calculated. This provides new insights on how different regulatory factors exert their action on bone cells leading to cellular spatial and temporal segregation, and functional coordination.     

A Data-Driven Random Subfeature Ensemble Learning Algorithm for Weather Forecasting

In this paper, the RSEL (Random Subfeature Ensemble Learning) algorithm is proposed to improve the forecast results of weather forecasting. Based on the classical machine learning algorithms, RSEL algorithm integrates random subfeature selection and ensemble learning combination strategy to enhance the diversity of the features and avoid the influence of a small number of unstable outliers generated randomly. Furthermore, the feature engineering schemes are designed for the weather forecast data to make full use of spatial or temporal context. RSEL algorithm is tested by forecasting the wind speed and direction, and it improves the forecast accuracy of traditional methods and has good robustness.     

Computational Aspects of Multiscale Simulation with the Lumped Particle Framework

First introduced in [2], the lumped particle framework is a flexible and numerically efficient framework for the modelling of particle transport in fluid flow. In this paper, the framework is expanded to simulate multicomponent particle-laden fluid flow. This is accomplished by introducing simulation protocols to model particles over a wide range of length and time scales. Consequently, we present a time ordering scheme and an approximate approach for accelerating the computation of evolution of different particle constituents with large differences in physical scales. We apply the extended framework on the temporal evolution of three particle constituents in sand-laden flow, and horizontal release of spherical particles. Furthermore, we evaluate the numerical error of the lumped particle model. In this context, we discuss the Velocity-Verlet numerical scheme, and show how to apply this to solving Newton's equations within the framework. We show that the increased accuracy of the Velocity-Verlet scheme is not lost when applied to the lumped particle framework.     

Approximation of Spatio-Temporal Random Processes Using Tensor Decomposition

A new representation of spatio-temporal random processes is proposed in this work. In practical applications, such processes are used to model velocity fields, temperature distributions, response of vibrating systems, to name a few. Finding an efficient representation for any random process leads to encapsulation of information which makes it more convenient for a practical implementations, for instance, in a computational mechanics problem. For a single-parameter process such as spatial or temporal process, the eigenvalue decomposition of the covariance matrix leads to the well-known Karhunen-Loève (KL) decomposition. However, for multiparameter processes such as a spatio-temporal process, the covariance function itself can be defined in multiple ways. Here the process is assumed to be measured at a finite set of spatial locations and a finite number of time instants. Then the spatial covariance matrices at different time instants are considered to define the covariance of the process. This set of square, symmetric, positive semi-definite matrices is then represented as a third-order tensor. A suitable decomposition of this tensor can identify the dominant components of the process, and these components are then used to define a closed-form representation of the process. The procedure is analogous to the KL decomposition for a single-parameter process, however, the decompositions and interpretations vary significantly. The tensor decompositions are successfully applied to (i) a heat conduction problem, (ii) a vibration problem, and (iii) a covariance function taken from the literature that was fitted to model a measured wind velocity data. It is observed that the proposed representation provides an efficient approximation to some processes. Furthermore, a comparison with KL decomposition showed that the proposed method is computationally cheaper than the KL, both in terms of computer memory and execution time.     

Prediction for global African swine fever outbreaks based on a combination of random forest algorithms and meteorological data

African swine fever (ASF) is a virulent infectious disease of pigs. As there is no effective vaccine and treatment method at present, it poses a great threat to the pig industry once it breaks out. In this paper, we used ASF outbreak data and the WorldClim database meteorological data and selected the CfsSubset Evaluator‐Best First feature selection method combined with the random forest algorithms to construct an African swine fever outbreak prediction model. Subsequently, we also established a test set for data other than modelling, and the accuracy accuracy value range of the model on the independent test set was 76.02%–84.64%, which indicated that the modelling effect was better and the prediction accuracy was higher than previous estimates. In addition, logistic regression analysis was conducted on 12 features used for modelling and the ROC curves were drawn. The results showed that the bio14 features (precipitation of driest month) had the largest contribution to the outbreak of ASF, and it was speculated that the outbreak of the epidemic was significantly related to precipitation. Finally, we used this qualitative prediction model to build a global online prediction system for ASF outbreaks, in the hope that this study will help to decision‐makers who can then take the relevant prevention and control measures in order to prevent the further spread of future epidemics of the disease.     

Bilateral generic working memory circuit requires left-lateralized addition for verbal processing

According to the Baddeley-Hitch model, phonological and visuospatial representations are separable components of working memory (WM) linked by a central executive. The traditional view that the separation reflects the relative contribution of the 2 hemispheres (verbal WM--left; spatial WM--right) has been challenged by the position that a common bilateral frontoparietal network subserves both domains. Here, we test the hypothesis that there is a generic WM circuit that recruits additional specialized regions for verbal and spatial processing. We designed a functional magnetic resonance imaging paradigm to elicit activation in the WM circuit for verbal and spatial information using identical stimuli and applied this in 33 healthy controls. We detected left-lateralized quantitative differences in the left frontal and temporal lobe for verbal > spatial WM but no areas of activation for spatial > verbal WM. We speculate that spatial WM is analogous to a "generic" bilateral frontoparietal WM circuit we inherited from our great ape ancestors that evolved, by recruitment of additional left-lateralized frontal and temporal regions, to accommodate language.     

基于多层次集成学习的骨质疏松辅助诊断研究

目的 原发性骨质疏松是一种起病隐匿、病程较长,在中老年人中高发的疾病,其可引起包括骨折在内的一系列严重症状,是我国中老年人致残致死的主要原因之一。与骨质疏松相关的生理检验指标有很多,如何筛选利用这些指标为诊断服务、建立诊断模型,尚未有成熟、统一的方法。方法 利用人工智能相关技术,对临床骨质疏松患者指标使用多种特征相关性算法进行特征选择,并在此基础上提出了一种多层次的集成学习框架：SAB-SVMKNN算法,其通过将内部同质学习器集成和外部异质学习器集成结合,将集成学习中的Boosting算法和Bagging算法使用Stacking进行集成,构建性能更强,适应性更好地诊断预测模型。结果 使用特征选择从原始数据中的31项临床指标中筛选了对于骨质疏松最重要的8种相关特征,通过这种方式使各模型准确率平均提高了9.2%,且该研究对应的模型准确率提升18.6%,最终达到了94.8%的准确率。结论 特征选择对于临床诊断和骨质疏松疾病的研究具有重要意义,该研究构建的预测模型可以有助于提高医生的诊断准确率。     

Short-term relationship between meteorological variables and hip fractures: An analysis carried out in a health area of the Autonomous Region of Valencia, Spain (1996–2005)

IntroductionDiverse studies have shown a seasonal influence on the incidence of hip fracture (HF), possibly associated with adverse meteorological conditions. In this paper, we present an analysis of the short-term relationship between meteorological conditions and the incidence of HF in people over 45 years of age living in a Mediterranean climate zone.Material and methodsHF cases admitted to the two reference hospitals in the health area were selected. The meteorological variables (temperature, relative humidity, rain, wind, and other conditions) were obtained from a weather station centrally located within the area under study. The trend and seasonality of the time series were analyzed with the aid of Poisson regression modeling. The relationship between the incidence of a hip fracture and the meteorological conditions, both on the same day and on the day previous to the patient's admission to the hospital were correlated in a case-crossover analysis with the control periods selected in accordance with two different methods of approximation (symmetric and semi-metric). The results were analyzed for different subgroups defined by age (older or younger than 75 years of age) and sex of subject and by type of fracture (cervical or pertrochanteric).Results2121 patients admitted for HF were selected. Of these, 1598 (75.3%) were women and 523 (24.7%) were men. The average age of the subjects was 80, with patients ranging from 45–99 years of age. The time/weather series showed a positive tendency, with a greater occurrence of cases in the autumn and winter months. The case-crossover analysis showed a significant relationship between the daily duration of wind and the incidence of HFs. Divided into quartiles, the windiest days (quartile 4) were associated with a 32% increased risk of HF (OR 1.32 CI 95% 1.10–1.58) with respect to the calmest days (quartile 1), especially in patients under 75 (OR 1.53; CI 95% 1.02–2.29). The remaining meteorological variables were not associated in any significant fashion with the incidence of HFs. The results were comparable across different subgroups classified by age, sex, and type of fracture.ConclusionsThe incidence of HFs varies seasonally and presents a significant association with the coldest times of the year. In the short-term, lasting windy periods are associated with an increased risk of HFs whereas other meteorological variables show a much lower, non-significant correlation.     

Differential short-term synaptic plasticity and transmission of complex spike trains: to depress or to facilitate?

Experimental studies have revealed conspicuous short-term facilitation and depression that are expressed differentially at distinct classes of cortical synapses. To explore computational implications of synaptic dynamics, we investigated transmission of complex spike trains through a stochastic model of cortical synapse endowed with short-term facilitation and vesicle depletion. Inputs to the synapse model were either real spike train data recorded from the visual and prefrontal cortices of behaving monkeys, or were generated numerically with prescribed temporal statistics. We tested the hypothesis that short-term facilitation could enable synapses to filter out single spikes and favor bursts of action potentials. We found that the ratio between release probabilities for a burst spike and an isolated spike grows monotonically with increasing number of spikes per burst, and with increasing interval between isolated spikes. Burst detection is optimal when the facilitation time constant matches the average burst duration. Using fractal-like spike patterns characterized by long-term power-law temporal correlations and similar to those seen in sensory neurons, we found that facilitation increases correlation at short time scales. In contrast, depression leads to a dramatic reduction in temporal correlations at all time scales, and to a flat ('whitened') power spectrum, thereby decorrelating natural input signals.     

Space-Time Discontinuous Galerkin Method for Maxwell's Equations

A fully discrete discontinuous Galerkin method is introduced for solving time-dependent Maxwell's equations. Distinguished from the Runge-Kutta discontinuous Galerkin method (RKDG) and the finite element time domain method (FETD), in our scheme, discontinuous Galerkin methods are used to discretize not only the spatial domain but also the temporal domain. The proposed numerical scheme is proved to be unconditionally stable, and a convergent rate $\mathcal{O}((∆t)^{r+1}+h^{k+1/2})$ is established under the $L^2$ -norm when polynomials of degree at most $r$ and $k$ are used for temporal and spatial approximation, respectively. Numerical results in both 2-D and 3-D are provided to validate the theoretical prediction. An ultra-convergence of order $(∆t)^{2r+1}$ in time step is observed numerically for the numerical fluxes w.r.t. temporal variable at the grid points.     

直肠癌与腹膜反折位置相关的MRI表现分析

目的:评价MRI检查判断直肠癌肿块与腹膜反折位置关系的准确性。方法:2010年1月至2012年12月我院64例直肠癌手术病人,术前行薄层MRI检查,由两位高年资医师行影像评估,并与术中发现对照。结果:本研究64例病人均经手术证实,常规序列术前MRI检查直肠癌位置的准确率为90.6%。MRI判断直肠癌位于腹膜反折之上的诊断符合率为94.4%,直肠癌位于腹膜反折处的诊断符合率为84.6%,直肠癌位于腹膜反折之下的诊断符合率为86.7%。结论:MRI检查可用于判断直肠癌与腹膜反折的位置关系,指导制定正确的治疗方案。     

Large-scale neural model for visual attention: integration of experimental single-cell and fMRI data

A computational neuroscience framework is proposed to better understand the role and the neuronal correlate of spatial attention modulation in visual perception. The model consists of several interconnected modules that can be related to the different areas of the dorsal and ventral paths of the visual cortex. Competitive neural interactions are implemented at both microscopic and interareal levels, according to the biased competition hypothesis. This hypothesis has been experimentally confirmed in studies in humans using functional magnetic resonance imaging (fMRI) techniques and also in single-cell recording studies in monkeys. Within this neuro-dynamical approach, numerical simulations are carried out that describe both the fMRI and the electrophysiological data. The proposed model draws together data of different spatial and temporal resolution, as are the above-mentioned imaging and single-cell results.     

Magnetic source imaging in stereotactic and functional neurosurgery

Magnetic source imaging (MSI) combines the unique spatial and temporal functional accuracy of magnetoencephalography (MEG) with the anatomic and pathologic detail of magnetic resonance (MR). This relatively new method of evaluating brain function provides a preoperative mapping of brain function and brain structure by integrating the functional information of MEG with the structural information of MR. This results in data on actual neuronal interactions in clinical patients. The temporal and spatial accuracy of the MEG data, combined with the anatomic and pathologic specificity of MRI, results in the magnetic source image, which offers accurate knowledge of cortical functional organization, and is important in the surgical treatment of brain neoplasms, vascular malformations, and epilepsy. MSI allows the tracking of neuronal activity on the scale of milliseconds with millimeter accuracy, and continues to lead to new understanding of many functional brain disorders.     

Famous faces activate contextual associations in the parahippocampal cortex

The parahippocampal cortex (PHC) has been traditionally implicated both in place processing and in episodic memory. How could the same cortical region mediate these cognitive functions that seem quite different? We have recently proposed that the PHC should be seen as more generally mediating contextual associative processing, which is required for both navigation and memory. We therefore predicted that any associative objects should activate the PHC. To test this generalization, we investigated the extent to which common stimuli that are nonspatial by nature, namely faces, activate the PHC, although their perception is typically associated with other cortical structures. Specifically, we compared the activation elicited by famous faces, which are highly associated with rich pictorial and contextual information (e.g., Tom Cruise) and are not associated with a specific place, with activation elicited by unfamiliar faces. Consistent with our prediction, contrasting famous with unfamiliar faces revealed significant activation within the PHC. Taken collectively, these findings indicate that the PHC should be regarded as mediating contextual associations in general and not necessarily spatial or episodic information.     

ODI及SBI指数、SF-36量表预测脊柱复发性椎间盘突出症再手术患者预后的研究

目的:探讨ODI、SBI及SF 36量表对脊柱复发性椎间盘突出症再手术患者预后的预测价值。方法 :选取2013年6月至2015年12月因复发性椎间盘突出症再次行手术治疗的患者,按照临床疗效好中差依次分为A、B、C3组,并按70∶30比例分为训练集和测试集,采用有序Logistic回归构建预测模型,并用测试集对模型效果进行验证,计算模型准确率。结果:A、B两组患者ODI与SBI分值均较C组更低,而两组患者SF 36量表所得分值显著高于C组(P0.05);采用有序Logistic回归构建预测疗效模型得出ODI系数为0.67,SF 36系数为-0.43、SBI系数为0.52;在临床疗效好的A组,模型预测正确率为80.00%,在临床疗效为中的B组,模型预测正确率为76.92%;在临床疗效为差的C组,模型预测正确率为44.44%。结论:综合考虑ODI、SBI与SF 36构建复发性椎间盘突出症再手术后患者的临床疗效预测模型,能够较好地预测脊柱复发性椎间盘突出症再手术患者的预后,具有一定的临床应用价值。     

Saturation-Value Blind Color Image Deblurring with Geometric Spatial-Feature Prior

Blind deblurring for color images has long been a challenging computer vision task. The intrinsic color structures within image channels have typically been disregarded in many excellent works. We investigate employing regularizations in the hue, saturation, and value (HSV) color space via the quaternion framework in order to better retain the internal relationship among the multiple channels and reduce color distortions and color artifacts. We observe that a geometric spatial-feature prior utilized in the intermediate latent image successfully enhances the kernel accuracy for the blind deblurring variational models, preserving the salient edges while decreasing the unfavorable structures. Motivated by this, we develop a saturation-value geometric spatial-feature prior in the HSV color space via the quaternion framework for blind color image deblurring, which facilitates blur kernel estimation. An alternating optimization strategy combined with a primal-dual projected gradient method can effectively solve this novel proposed model. Extensive experimental results show that our model outperforms state-of-the-art methods in blind color image deblurring by a wide margin, demonstrating the effectiveness of the proposed model.     

Repositioning accuracy and movement parameters in low back pain subjects and healthy control subjects

Study Design: A control group study with repeated measures. Objective: To compare trunk repositioning parameters in chronic low back pain (LBP) and healthy subjects. Summary and background data: Recent evidence suggests that chronic LBP patients exhibit deficits in trunk proprioception and motor control. Trunk repositioning and the various spatio-temporal parameters related to it can be used to evaluate sensori-motor control and movement strategies. Methods: Fifteen control subjects and 16 chronic LBP subjects participated in this study. Subjects were required to reproduce different trunk position in flexion (15°, 30° and 60°) and extension (15°). In the learning phase preceding each condition, visual feedback was provided. Following these learning trials, subjects were asked to perform ten consecutive trials without any feedback. Movement time, movement time variability and peak velocity were obtained and a temporal symmetry ratio was calculated. Peak angular position variability and absolute error in peak angular position were also calculated to evaluate spatial accuracy. Results: Two subgroups of LBP patients were identified. One subgroup of LBP subjects demonstrated longer movement time and smaller peak velocities and symmetry ratios than normal subjects. No group difference was observed for peak angular position variability and absolute error in peak angular position. Conclusion: Chronic LBP patients, when given a sufficient learning period, were able to reproduce trunk position with a spatial accuracy similar to control subjects. Some LBP subjects, however, showed modifications of movement time, peak velocity and acceleration parameters. We propose that the presence of persistent chronic pain could induce an alteration or an adaptation in the motor responses of chronic LBP subjects.     

Conservative Semi-Lagrangian Finite Difference WENO Formulations with Applications to the Vlasov Equation

In this paper, we propose a new conservative semi-Lagrangian (SL) finite difference (FD) WENO scheme for linear advection equations, which can serve as a base scheme for the Vlasov equation by Strang splitting [4]. The reconstruction procedure in the proposed SL FD scheme is the same as the one used in the SL finite volume (FV) WENO scheme [3]. However, instead of inputting cell averages and approximate the integral form of the equation in a FV scheme, we input point values and approximate the differential form of equation in a FD spirit, yet retaining very high order (fifth order in our experiment) spatial accuracy. The advantage of using point values, rather than cell averages, is to avoid the second order spatial error, due to the shearing in velocity (v) and electrical field (E) over a cell when performing the Strang splitting to the Vlasov equation. As a result, the proposed scheme has very high spatial accuracy, compared with second order spatial accuracy for Strang split SL FV scheme for solving the Vlasov-Poisson (VP) system. We perform numerical experiments on linear advection, rigid body rotation problem; and on the Landau damping and two-stream instabilities by solving the VP system. For comparison, we also apply (1) the conservative SL FD WENO scheme, proposed in [22] for incompressible advection problem, (2) the conservative SL FD WENO scheme proposed in [21] and (3) the non-conservative version of the SL FD WENO scheme in [3] to the same test problems. The performances of different schemes are compared by the error table, solution resolution of sharp interface, and by tracking the conservation of physical norms, energies and entropies, which should be physically preserved.