Causality-driven multivariate stock movement forecasting.

Our study aims to investigate the interdependence between international stock markets and sentiments from financial news in stock forecasting. We adopt the Temporal Fusion Transformers (TFT) to incorporate intra and inter-market correlations and the interaction between the information flow, i.e. causality, of financial news sentiment and the dynamics of the stock market. The current study distinguishes itself from existing research by adopting Dynamic Transfer Entropy (DTE) to establish an accurate information flow propagation between stock and sentiments. DTE has the advantage of providing time series that mine information flow propagation paths between certain parts of the time series, highlighting marginal events such as spikes or sudden jumps, which are crucial in financial time series. The proposed methodological approach involves the following elements: a FinBERT-based textual analysis of financial news articles to extract sentiment time series, the use of the Transfer Entropy and corresponding heat maps to analyze the net information flows, the calculation of the DTE time series, which are considered as co-occurring covariates of stock Price, and TFT-based stock forecasting. The Dow Jones Industrial Average index of 13 countries, along with daily financial news data obtained through the New York Times API, are used to demonstrate the validity and superiority of the proposed DTE-based causality method along with TFT for accurate stock Price and Return forecasting compared to state-of-the-art time series forecasting methods.

Long-term Regional Influenza-like-illness Forecasting Using Exogenous Data.

Disease forecasting is a longstanding problem for the research community, which aims at informing and improving decisions with the best available evidence. Specifically, the interest in respiratory disease forecasting has dramatically increased since the beginning of the coronavirus pandemic, rendering the accurate prediction of influenza-like-illness (ILI) a critical task. Although methods for short-term ILI forecasting and nowcasting have achieved good accuracy, their performance worsens at long-term ILI forecasts. Machine learning models have outperformed conventional forecasting approaches enabling to utilize diverse exogenous data sources, such as social media, internet users' search query logs, and climate data. However, the most recent deep learning ILI forecasting models use only historical occurrence data achieving state-of-the-art results. Inspired by recent deep neural network architectures in time series forecasting, this work proposes the Regional Influenza-Like-Illness Forecasting (ReILIF) method for regional long-term ILI prediction. The proposed architecture takes advantage of diverse exogenous data, that are, meteorological and population data, introducing an efficient intermediate fusion mechanism to combine the different types of information with the aim to capture the variations of ILI from various views. The efficacy of the proposed approach compared to state-of-the-art ILI forecasting methods is confirmed by an extensive experimental study following standard evaluation measures.

Semi-supervised medical image classification via distance correlation minimization and graph attention regularization.

We propose a novel semi-supervised learning method to leverage unlabeled data alongside minimal annotated data and improve medical imaging classification performance in realistic scenarios with limited labeling budgets to afford data annotations. Our method introduces distance correlation to minimize correlations between feature representations from different views of the same image encoded with non-coupled deep neural networks architectures. In addition, it incorporates a data-driven graph-attention based regularization strategy to model affinities among images within the unlabeled data by exploiting their inherent relational information in the feature space. We conduct extensive experiments on four medical imaging benchmark data sets involving X-ray, dermoscopic, magnetic resonance, and computer tomography imaging on single and multi-label medical imaging classification scenarios. Our experiments demonstrate the effectiveness of our method in achieving very competitive performance and outperforming several state-of-the-art semi-supervised learning methods. Furthermore, they confirm the suitability of distance correlation as a versatile dependence measure and the benefits of the proposed graph-attention based regularization for semi-supervised learning in medical imaging analysis.

Rayleigh-based segmentation of ISAR images.

Inverse synthetic aperture radar (ISAR) provides a solution to increase the radar angular resolution by observing a moving target over time. The high-resolution ISAR image should undergo a segmentation step to get the target's point cloud data, which is then used for classification purposes. Existing segmentation algorithms seek an optimal threshold in an iterative manner, which adds to the complexity of ISAR and results in an increase in the processing time. In this paper, we take advantage of the distribution of the ISAR image intensity, which is based on the Rayleigh distribution, and obtain an explicit relationship for the optimal segmentation threshold. The proposed segmentation algorithm alleviates the requirement for iterative optimization and its efficiency is shown using both simulated and experimental ISAR images.

A multidimensional ODE-based model of Alzheimer's disease progression.

Data-driven Alzheimer's disease (AD) progression models are useful for clinical prediction, disease mechanism understanding, and clinical trial design. Most dynamic models were inspired by the amyloid cascade hypothesis and described AD progression as a linear chain of pathological events. However, the heterogeneity observed in healthy and sporadic AD populations challenged the amyloid hypothesis, and there is a need for more flexible dynamical models that accompany this conceptual shift. We present a statistical model of the temporal evolution of biomarkers and cognitive tests that allows diverse biomarker paths throughout the disease. The model consists of two elements: a multivariate dynamic model of the joint evolution of biomarkers and cognitive tests; and a clinical prediction model. The dynamic model uses a system of ordinary differential equations to jointly model the rate of change of an individual's biomarkers and cognitive tests. The clinical prediction model is an ordinal logistic model of the diagnostic label. Prognosis and time-to-onset predictions are obtained by computing the clinical label probabilities throughout the forecasted biomarker trajectories. The proposed dynamical model is interpretable, free of one-dimensional progression hypotheses or disease staging paradigms, and can account for the heterogeneous dynamics observed in sporadic AD. We developed the model using longitudinal data from the Alzheimer's Disease Neuroimaging Initiative. We illustrate the patterns of biomarker rates of change and the model performance to predict the time to conversion from MCI to dementia.

Beyond shallow feelings of complex affect: Non-motor correlates of subjective emotional experience in Parkinson's disease.

Affective disorders in Parkinson's disease (PD) concern several components of emotion. However, research on subjective feeling in PD is scarce and has produced overall varying results. Therefore, in this study, we aimed to evaluate the subjective emotional experience and its relationship with autonomic symptoms and other non-motor features in PD patients. We used a battery of film excerpts to elicit Amusement, Anger, Disgust, Fear, Sadness, Tenderness, and Neutral State, in 28 PD patients and 17 healthy controls. Self-report scores of emotion category, intensity, and valence were analyzed. In the PD group, we explored the association between emotional self-reported scores and clinical scales assessing autonomic dysregulation, depression, REM sleep behavior disorder, and cognitive impairment. Patient clustering was assessed by considering relevant associations. Tenderness occurrence and intensity of Tenderness and Amusement were reduced in the PD patients. Tenderness occurrence was mainly associated with the overall cognitive status and the prevalence of gastrointestinal symptoms. In contrast, the intensity and valence reported for the experience of Amusement correlated with the prevalence of urinary symptoms. We identified five patient clusters, which differed significantly in their profile of non-motor symptoms and subjective feeling. Our findings further suggest the possible existence of a PD phenotype with more significant changes in subjective emotional experience. We concluded that the subjective experience of complex emotions is impaired in PD. Non-motor feature grouping suggests the existence of disease phenotypes profiled according to specific deficits in subjective emotional experience, with potential clinical implications for the adoption of precision medicine in PD. Further research on larger sample sizes, combining subjective and physiological measures of emotion with additional clinical features, is needed to extend our findings.

Stable Sparse Classifiers predict cognitive impairment from gait patterns.

Background: Although gait patterns disturbances are known to be related to cognitive decline, there is no consensus on the possibility of predicting one from the other. It is necessary to find the optimal gait features, experimental protocols, and computational algorithms to achieve this purpose. Purposes: To assess the efficacy of the Stable Sparse Classifiers procedure (SSC) for discriminating young and healthy older adults (YA vs. HE), as well as healthy and cognitively impaired elderly groups (HE vs. MCI-E) from their gait patterns. To identify the walking tasks or combinations of tasks and specific spatio-temporal gait features (STGF) that allow the best prediction with SSC. Methods: A sample of 125 participants (40 young- and 85 older-adults) was studied. They underwent assessment with five neuropsychological tests that explore different cognitive domains. A summarized cognitive index (MDCog), based on the Mahalanobis distance from normative data, was calculated. The sample was divided into three groups (young adults, healthy and cognitively impaired elderly adults) using k-means clustering of MDCog in addition to Age. The participants executed four walking tasks (normal, fast, easy- and hard-dual tasks) and their gait patterns, measured with a body-fixed Inertial Measurement Unit, were used to calculate 16 STGF and dual-task costs. SSC was then employed to predict which group the participants belonged to. The classification's performance was assessed using the area under the receiver operating curves (AUC) and the stable biomarkers were identified. Results: The discrimination HE vs. MCI-E revealed that the combination of the easy dual-task and the fast walking task had the best prediction performance (AUC = 0.86, sensitivity: 90.1%, specificity: 96.9%, accuracy: 95.8%). The features related to gait variability and to the amplitude of vertical acceleration had the largest predictive power. SSC prediction accuracy was better than the accuracies obtained with linear discriminant analysis and support vector machine classifiers. Conclusions: The study corroborated that the changes in gait patterns can be used to discriminate between young and healthy older adults and more importantly between healthy and cognitively impaired adults. A subset of gait tasks and STGF optimal for achieving this goal with SSC were identified, with the latter method superior to other classification techniques.

Machine Learning Techniques for the Diagnosis of Schizophrenia Based on Event-Related Potentials.

Antecedent: The event-related potential (ERP) components P300 and mismatch negativity (MMN) have been linked to cognitive deficits in patients with schizophrenia. The diagnosis of schizophrenia could be improved by applying machine learning procedures to these objective neurophysiological biomarkers. Several studies have attempted to achieve this goal, but no study has examined Multiple Kernel Learning (MKL) classifiers. This algorithm finds optimally a combination of kernel functions, integrating them in a meaningful manner, and thus could improve diagnosis. Objective: This study aimed to examine the efficacy of the MKL classifier and the Boruta feature selection method for schizophrenia patients (SZ) and healthy controls (HC) single-subject classification. Methods: A cohort of 54 SZ and 54 HC participants were studied. Three sets of features related to ERP signals were calculated as follows: peak related features, peak to peak related features, and signal related features. The Boruta algorithm was used to evaluate the impact of feature selection on classification performance. An MKL algorithm was applied to address schizophrenia detection. Results: A classification accuracy of 83% using the whole dataset, and 86% after applying Boruta feature selection was obtained. The variables that contributed most to the classification were mainly related to the latency and amplitude of the auditory P300 paradigm. Conclusion: This study showed that MKL can be useful in distinguishing between schizophrenic patients and controls when using ERP measures. Moreover, the use of the Boruta algorithm provides an improvement in classification accuracy and computational cost.

Statistical parametrization of cell cytoskeleton reveals lung cancer cytoskeletal phenotype with partial EMT signature.

Epithelial-mesenchymal Transition (EMT) is a multi-step process that involves cytoskeletal rearrangement. Here, developing and using an image quantification tool, Statistical Parametrization of Cell Cytoskeleton (SPOCC), we have identified an intermediate EMT state with a specific cytoskeletal signature. We have been able to partition EMT into two steps: (1) initial formation of transverse arcs and dorsal stress fibers and (2) their subsequent conversion to ventral stress fibers with a concurrent alignment of fibers. Using the Orientational Order Parameter (OOP) as a figure of merit, we have been able to track EMT progression in live cells as well as characterize and quantify their cytoskeletal response to drugs. SPOCC has improved throughput and is non-destructive, making it a viable candidate for studying a broad range of biological processes. Further, owing to the increased stiffness (and by inference invasiveness) of the intermediate EMT phenotype compared to mesenchymal cells, our work can be instrumental in aiding the search for future treatment strategies that combat metastasis by specifically targeting the fiber alignment process.

Single yeast cell nanomotions correlate with cellular activity.

Living single yeast cells show a specific cellular motion at the nanometer scale with a magnitude that is proportional to the cellular activity of the cell. We characterized this cellular nanomotion pattern of nonattached single yeast cells using classical optical microscopy. The distribution of the cellular displacements over a short time period is distinct from random motion. The range and shape of such nanomotion displacement distributions change substantially according to the metabolic state of the cell. The analysis of the nanomotion frequency pattern demonstrated that single living yeast cells oscillate at relatively low frequencies of around 2 hertz. The simplicity of the technique should open the way to numerous applications among which antifungal susceptibility tests seem the most straightforward.

SVRG-MKL: A Fast and Scalable Multiple Kernel Learning Solution for Features Combination in Multi-Class Classification Problems.

In this paper, we present a novel strategy to combine a set of compact descriptors to leverage an associated recognition task. We formulate the problem from a multiple kernel learning (MKL) perspective and solve it following a stochastic variance reduced gradient (SVRG) approach to address its scalability, currently an open issue. MKL models are ideal candidates to jointly learn the optimal combination of features along with its associated predictor. However, they are unable to scale beyond a dozen thousand of samples due to high computational and memory requirements, which severely limits their applicability. We propose SVRG-MKL, an MKL solution with inherent scalability properties that can optimally combine multiple descriptors involving millions of samples. Our solution takes place directly in the primal to avoid Gram matrices computation and memory allocation, whereas the optimization is performed with a proposed algorithm of linear complexity and hence computationally efficient. Our proposition builds upon recent progress in SVRG with the distinction that each kernel is treated differently during optimization, which results in a faster convergence than applying off-the-shelf SVRG into MKL. Extensive experimental validation conducted on several benchmarking data sets confirms a higher accuracy and a significant speedup of our solution. Our technique can be extended to other MKL problems, including visual search and transfer learning, as well as other formulations, such as group-sensitive (GMKL) and localized MKL (LMKL) in convex settings.

A Robust Actin Filaments Image Analysis Framework.

The cytoskeleton is a highly dynamical protein network that plays a central role in numerous cellular physiological processes, and is traditionally divided into three components according to its chemical composition, i.e. actin, tubulin and intermediate filament cytoskeletons. Understanding the cytoskeleton dynamics is of prime importance to unveil mechanisms involved in cell adaptation to any stress type. Fluorescence imaging of cytoskeleton structures allows analyzing the impact of mechanical stimulation in the cytoskeleton, but it also imposes additional challenges in the image processing stage, such as the presence of imaging-related artifacts and heavy blurring introduced by (high-throughput) automated scans. However, although there exists a considerable number of image-based analytical tools to address the image processing and analysis, most of them are unfit to cope with the aforementioned challenges. Filamentous structures in images can be considered as a piecewise composition of quasi-straight segments (at least in some finer or coarser scale). Based on this observation, we propose a three-steps actin filaments extraction methodology: (i) first the input image is decomposed into a 'cartoon' part corresponding to the filament structures in the image, and a noise/texture part, (ii) on the 'cartoon' image, we apply a multi-scale line detector coupled with a (iii) quasi-straight filaments merging algorithm for fiber extraction. The proposed robust actin filaments image analysis framework allows extracting individual filaments in the presence of noise, artifacts and heavy blurring. Moreover, it provides numerous parameters such as filaments orientation, position and length, useful for further analysis. Cell image decomposition is relatively under-exploited in biological images processing, and our study shows the benefits it provides when addressing such tasks. Experimental validation was conducted using publicly available datasets, and in osteoblasts grown in two different conditions: static (control) and fluid shear stress. The proposed methodology exhibited higher sensitivity values and similar accuracy compared to state-of-the-art methods.

Lectin-Glycan Interaction Network-Based Identification of Host Receptors of Microbial Pathogenic Adhesins.

UNLABELLED: The first step in the infection of humans by microbial pathogens is their adherence to host tissue cells, which is frequently based on the binding of carbohydrate-binding proteins (lectin-like adhesins) to human cell receptors that expose glycans. In only a few cases have the human receptors of pathogenic adhesins been described. A novel strategy-based on the construction of a lectin-glycan interaction (LGI) network-to identify the potential human binding receptors for pathogenic adhesins with lectin activity was developed. The new approach is based on linking glycan array screening results of these adhesins to a human glycoprotein database via the construction of an LGI network. This strategy was used to detect human receptors for virulent Escherichia coli (FimH adhesin), and the fungal pathogens Candida albicans (Als1p and Als3p adhesins) and C. glabrata (Epa1, Epa6, and Epa7 adhesins), which cause candidiasis. This LGI network strategy allows the profiling of potential adhesin binding receptors in the host with prioritization, based on experimental binding data, of the most relevant interactions. New potential targets for the selected adhesins were predicted and experimentally confirmed. This methodology was also used to predict lectin interactions with envelope glycoproteins of human-pathogenic viruses. It was shown that this strategy was successful in revealing that the FimH adhesin has anti-HIV activity. IMPORTANCE: Microbial pathogens may express a wide range of carbohydrate-specific adhesion proteins that mediate adherence to host tissues. Pathogen attachment to host cells is achieved through the binding of these lectin-like adhesins to glycans on human glycoproteins. In only a few cases have the human receptors of pathogenic adhesins been described. We developed a new strategy to predict these interacting receptors. Therefore, we developed a novel LGI network that would allow the mapping of potential adhesin binding receptors in the host with prioritization, based on the experimental binding data, of the most relevant interactions. New potential targets for the selected adhesins (bacterial uroepithelial FimH from E. coli and fungal Epa and Als adhesins from C. glabrata and C. albicans) were predicted and experimentally confirmed. This methodology was also used to predict lectin interactions with human-pathogenic viruses and to discover whether FimH adhesin has anti-HIV activity.

Objectifying facial expressivity assessment of Parkinson's patients: preliminary study.

Patients with Parkinson's disease (PD) can exhibit a reduction of spontaneous facial expression, designated as "facial masking," a symptom in which facial muscles become rigid. To improve clinical assessment of facial expressivity of PD, this work attempts to quantify the dynamic facial expressivity (facial activity) of PD by automatically recognizing facial action units (AUs) and estimating their intensity. Spontaneous facial expressivity was assessed by comparing 7 PD patients with 8 control participants. To voluntarily produce spontaneous facial expressions that resemble those typically triggered by emotions, six emotions (amusement, sadness, anger, disgust, surprise, and fear) were elicited using movie clips. During the movie clips, physiological signals (facial electromyography (EMG) and electrocardiogram (ECG)) and frontal face video of the participants were recorded. The participants were asked to report on their emotional states throughout the experiment. We first examined the effectiveness of the emotion manipulation by evaluating the participant's self-reports. Disgust-induced emotions were significantly higher than the other emotions. Thus we focused on the analysis of the recorded data during watching disgust movie clips. The proposed facial expressivity assessment approach captured differences in facial expressivity between PD patients and controls. Also differences between PD patients with different progression of Parkinson's disease have been observed.

The effect of material modeling on finite element analysis of human breast biomechanics.

PURPOSE: Finite element analysis has been used extensively in the study of biomechanical modeling of the breast. However, issues regarding the complexity of material models and the influences of geometric boundary conditions on the accuracy of a breast Finite Element (FE) model are still under debate. This work demonstrates the importance of material modeling in FE models of the breast. METHODS: A simple hemispherical geometry is used to model the shape of a human breast. Different material models are being investigated to accurately model changes in terms of displacement, stress, and reaction forces distribution. RESULTS: The results obtained using nonlinear material models are compared with those obtained employing their linear approximation. Results have shown that differences, in terms of displacement, ranging between 20% and more than 80%, may occur and that large differences are present in terms of maximum principal stresses when the displacement is correctly approximated. CONCLUSIONS: This study clearly shows that, in a FE model, simulating large deformations material modeling strongly influences the accuracy of the solution.

A high resolution bladder wall map: feasibility study.

The aim of this work is to provide the surgeon-urologist with a system for automatic 2D and 3D-reconstruction of the bladder wall to help him within the treatment of bladder cancer as well as planning and documentation of the interventions. Within this small pilot-framework a fast feasibility study was made to clear if it is generally possible to build a bladder wall model using a special endoscope with an embedded laser-based distance measurement, an optical navigation system and modern image stitching techniques. Some experiments with a realistic bladder phantom have shown that this initial concept is generally acceptable and can be used with some extensions to build a system which can provide an automatic bladder wall reconstruction in real time to be used within a surgical intervention.

Multiscale gradient watersheds of color images.

We present a new framework for the hierarchical segmentation of color images. The proposed scheme comprises a nonlinear scale-space with vector-valued gradient watersheds. Our aim is to produce a meaningful hierarchy among the objects in the image using three image components of distinct perceptual significance for a human observer, namely strong edges, smooth segments and detailed segments. The scale-space is based on a vector-valued diffusion that uses the Additive Operator Splitting numerical scheme. Furthermore, we introduce the principle of the dynamics of contours in scale-space that combines scale and contrast information. The performance of the proposed segmentation scheme is presented via experimental results obtained with a wide range of images including natural and artificial scenes.