Distinct Information Research Articles

Pre-gating and contextual attention gate — A new fusion method for multi-modal data tasks

Multi-modal representation learning has received significant attention across diverse research domains due to its ability to model a scenario comprehensively. Learning the cross-modal interactions is essential to combining multi-modal data into a joint representation. However, conventional cross-attention mechanisms can produce noisy and non-meaningful values in the absence of useful cross-modal interactions among input features, thereby introducing uncertainty into the feature representation. These factors have the potential to degrade the performance of downstream tasks. This paper introduces a novel Pre-gating and Contextual Attention Gate (PCAG) module for multi-modal learning comprising two gating mechanisms that operate at distinct information processing levels within the deep learning model. The first gate filters out interactions that lack informativeness for the downstream task, while the second gate reduces the uncertainty introduced by the cross-attention module. Experimental results on eight multi-modal classification tasks spanning various domains show that the multi-modal fusion model with PCAG outperforms state-of-the-art multi-modal fusion models. Additionally, we elucidate how PCAG effectively processes cross-modality interactions

Transcriptomic Expression of T2-Inflammation Genes in Peripheral Blood Mononuclear Cells and Longitudinal Clinical Outcomes in Asthma: Insights from the COREA Study.

Gene expression can provide distinct information compared to clinical biomarkers in the context of longitudinal clinical outcomes in asthma patients. This study examined the association between the gene expression levels of upstream (IL-25, IL-33, and TSLP) and downstream cytokines (IL-5, IL-4, and IL-13) in the T2 inflammatory pathway with a 12-month follow-up of exacerbation, lung function, and steroid use. Transcriptomic sequencing analysis was performed on peripheral blood mononuclear cells from 279 adult asthmatics. Survival analysis and linear mixed-effect models were used to investigate potential differences between the high-level and low-level gene expression groups and the clinical outcomes. Analysis was performed separately for the upstream, downstream, and all 6 cytokines. In general, T2 inflammatory cytokine gene expression showed a weak correlation with blood eosinophil counts (all r < 0.1) and clinical outcomes. Among moderate-to-severe eosinophilic asthma (MSEA) patients, individuals with elevated levels of downstream cytokines were at increased risk of time-to-first exacerbation (p = 0.044) and a greater increase of inhaled corticosteroid use over time (p = 0.002) compared to those with lower gene expression. There was no association between baseline T2 inflammatory cytokine gene expression and the longitudinal changes in lung function over time among MSEA patients. These findings suggest that, among MSEA patients, the gene expression levels of downstream cytokines in the T2 inflammatory pathway may serve as indicators for endotyping asthma.

Anchor-guided global view reconstruction for multi-view multi-label feature selection

In multi-view multi-label learning (MVML), the accuracy of feature weights is pivotal for establishing feature order. However, conventional MVML methods often struggle with integrating distinct information from multiple views effectively, leading to unclear segmentation and potential noise introduction. To address this challenge, this paper proposes an anchor-based latent representation method for global view learning in MVML. Specifically, we encode inherent information from each view to derive a candidate multi-view representation. Anchors extracted from both the candidate view and the global view are then constrained to approximate equality in the latent space. Furthermore, a carefully designed view matrix serves as supplement, seamlessly integrated into the reconstruction process to augment information. The convergence of results is subsequently validated using multiplicative update rules. Experimental findings showcase the superior performance of our proposed method across various multi-view datasets.

Altered correlation of concurrently recorded EEG-fMRI connectomes in temporal lobe epilepsy.

Whole-brain functional connectivity networks (connectomes) have been characterized at different scales in humans using EEG and fMRI. Multimodal epileptic networks have also been investigated, but the relationship between EEG and fMRI defined networks on a whole-brain scale is unclear. A unified multimodal connectome description, mapping healthy and pathological networks would close this knowledge gap. Here, we characterize the spatial correlation between the EEG and fMRI connectomes in right and left temporal lobe epilepsy (rTLE/lTLE). From two centers, we acquired resting-state concurrent EEG-fMRI of 35 healthy controls and 34 TLE patients. EEG-fMRI data was projected into the Desikan brain atlas, and functional connectomes from both modalities were correlated. EEG and fMRI connectomes were moderately correlated. This correlation was increased in rTLE when compared to controls for EEG-delta/theta/alpha/beta. Conversely, multimodal correlation in lTLE was decreased in respect to controls for EEG-beta. While the alteration was global in rTLE, in lTLE it was locally linked to the default mode network. The increased multimodal correlation in rTLE and decreased correlation in lTLE suggests a modality-specific lateralized differential reorganization in TLE, which needs to be considered when comparing results from different modalities. Each modality provides distinct information, highlighting the benefit of multimodal assessment in epilepsy.

Social fish use different submission signals depending on spatial position

Abstract Animals often possess multiple distinct threat signals which indicate their level of aggressive motivation or resource holding potential. In contrast, the diversity of submission signals is considerably lower. Still, some social species exhibit multiple distinct submission signals which could have different strategic functions or serve to enhance signal transmission. In the current study, we examine submission signalling in the daffodil cichlid (Neolamprologus pulcher), a cooperatively breeding fish which produces two distinct submission signals: a tail quiver, and a head-up posture. Observing interactions within stable social groups revealed that submission signal selection depends on the location of the interaction, with tail quivers being preferentially displayed near the substrate, and head-up postures being deployed when the interaction takes place further away from the substrate. Our findings suggest that variation in submission signalling in daffodil cichlids may serve to enhance signal transmission across spatial contexts rather than to convey distinct information.

Identifying the Impacts of Access to Different Information on Emergency Behaviors During COVID-19 Pandemic

Existing studies have verified the effect of information access on health behaviors, while less is known about the different impacts of distinct information access in a pandemic context. This study identified four information types (including disease knowledge, dynamic news, government measures, and social news) and then investigated how different information access affected preventive behaviors via perceived risk and media credibility during the COVID-19 outbreak in China. A nationally representative sampling survey on Chinese netizens was conducted ([Formula: see text]). It found that access to dynamic news, disease knowledge, and government measures positively predicted perceived severity, while only social news positively predicted perceived susceptibility. Access to disease knowledge, government measures, and social news were positively correlated with media credibility. The mediating roles of perceived risk and media credibility were proved. Besides, the statement of “cues-to-action” was also proved, which indicated that information access impacted behavioral changes more directly rather than through the mediators during COVID-19 pandemic. These findings contribute to a better understanding of how information access impacts public health behaviors in a pandemic context.

Crystallographic phase identifier of a convolutional self-attention neural network (CPICANN) on powder diffraction patterns.

Spectroscopic data, particularly diffraction data, are essential for materials characterization due to their comprehensive crystallographic information. The current crystallographic phase identification, however, is very time consuming. To address this challenge, we have developed a real-time crystallographic phase identifier based on a convolutional self-attention neural network (CPICANN). Trained on 692 190 simulated powder X-ray diffraction (XRD) patterns from 23 073 distinct inorganic crystallographic information files, CPICANN demonstrates superior phase-identification power. Single-phase identification on simulated XRD patterns yields 98.5 and 87.5% accuracies with and without elemental information, respectively, outperforming JADE software (68.2 and 38.7%, respectively). Bi-phase identification on simulated XRD patterns achieves 84.2 and 51.5% accuracies, respectively. In experimental settings, CPICANN achieves an 80% identification accuracy, surpassing JADE software (61%). Integration of CPICANN into XRD refinement software will significantly advance the cutting-edge technology in XRD materials characterization.

Multi-view clustering via dynamic unified bipartite graph learning

Multi-view clustering algorithms based on graph learning have the ability to extract the potential association between data samples, which has been a concern of many researchers in recent years. However, existing algorithms have two limitations: (1) they directly learn from the raw graph, which includes noise and outliers, and they construct the graph filter statically, biasing the clustering results; (2) during graph construction, they mainly use the information of a single structure and fail to fully extract the multi-granular structural information among the data. To address these issues, this paper proposes a novel multi-view clustering method via dynamic unified bipartite graph learning. Specifically, a learnable graph filter is first refined to dynamically filter the original data feature space, gradually filtering out the undesirable high-frequency noise and achieving a clustering-friendly smooth representation. Second, a unified bipartite graph is constructed by combining the multi-granular structural information of different views to better explore the distinct and common information of each view. In one framework, the dynamic filter and multi-granular structure information are combined to iteratively learn the unified bipartite graph. An efficient iterative algorithm is designed to decompose the objective function into small-scale subproblems for solving. Extensive experiments on benchmark datasets show the superiority of the proposed algorithm over several existing state-of-the-art multi-view clustering algorithms.

Using continuous relative phase and modified vector coding analyses to quantify spinal coordination and coordinative variability for healthy and chronic low back pain patients: An exploratory comparative analysis

Differences in coordination and coordinative variability are common in people with low back pain. While differences may relate to the different analyses used to quantify these metrics, the preferred approach remains unclear. We aimed to compare coordination and coordinative variability, in people with and without low back pain performing a lifting/lowering task, using continuous relative phase and vector coding procedures, and to identify which technique better detects group differences. Upper lumbar (T12-L3), lower lumbar (L3-S1), and hip angular kinematics were measured using electromagnetic motion capture during 10 crate lifting/lowering repetitions from adults with (n = 47) and without (n = 17) low back pain. Coordination and coordinative variability for the Hip-Lower Lumbar and Lower Lumbar-Upper Lumbar joint pairs were quantified using mean absolute relative phase and deviation phase (continuous relative phase), and coupling angle and coupling angle variability (vector coding), respectively. T-tests examined group differences in coordination and variability. Cohen’s d bootstrapping analyses identified the more sensitive technique for detecting group differences. Less in-phase and more variable behavior was observed in the low back pain group, mostly independent of joint pair and analytical technique (P < 0.05, Cohen’s d range = 0.61 to 1.33). Qualitatively, the low back group limited motion at the lower lumbar spine during lifting/lowering. Continuous relative phase was more sensitive in detecting group differences in coordinative variability, while vector coding was more sensitive towards differences in coordination. These procedures convey distinct information and have their respective merits. Researchers should consider the choice of analytical techniques based on their study objectives.

Plasmon-enhanced chiral absorption through electric dipole–electric quadrupole interaction

Enantioselective interactions of chiral molecules include distinct absorptions to opposite-handed circularly polarized light, known as chiral absorption. Traditionally, chiral absorption has been primarily attributed to electric dipole (ED) and magnetic dipole (MD) interaction with molecular chirality. However, this approach falls short for large molecules that support high-order multipolar components, such as electric quadrupole (EQ) moment. Here, we introduce a theoretical model to study the chiral absorption of large molecules in the presence of plasmonic nanostructures. This model considers both ED–MD interaction and ED–EQ interaction enhanced by a resonant structure. We numerically study such interactions of the chiral molecular solution in the vicinity of an achiral plasmonic nano-resonator. Our results show the distinct spectral information of the chiral medium on- and off-resonance of the resonator.

Distribution of peak frequency and omnipolar voltage in electrograms across the atrial body and thoracic veins in a normal heart.

The innovative peak frequency mapping facilitates the quantification of electrogram sharpness. However, reference values for normal atrial tissue are currently undefined. In this study, we explored the distribution of peak frequency and omnipolar peak-to-peak voltage (V-max) in a normal heart. Twenty-two patients with structurally normal heart were included. Either the right atrium (RA) and superior vena cava (SVC) or the left atrium (LA) and pulmonary veins (PVs) were mapped during sinus rhythm. In total, 13,654 points in the RA and 4143 points in the SVC from 15 patients and 4662 points in the LA and 2761 points in PVs from 7 patients were analyzed. The correlation between peak frequency and V-max was weak (R = 0.223). The median peak frequency was larger in the SVC than in the RA (441 [358-524] Hz vs. 358 [291-441] Hz, P < 0.0001) and in PVs than in the LA (346 [253-441] Hz vs. 323 [262-397] Hz, P < 0.0001). Conversely, the median V-max was smaller in the SVC than in the RA (1.96 [0.77-3.75] mV vs. 4.11 [2.10-6.83] mV, P < 0.0001) and in PVs than in the LA (1.16 [0.33-3.17] mV vs. 4.42 [2.63-6.84] mV, P < 0.0001). More than 95% of peak frequencies were > 174Hz in the RA and > 185Hz in the LA, and > 95% of V-maxes were > 0.52 and > 1.07mV in the RA and LA, respectively. Given the limited correlation between peak frequency and V-max, and recognizing their potential to provide distinct information, they can be used complementarily. Employing these parameters to extract varied insights can provide comprehensive understandings of tissue characteristics.

Mammographic Texture versus Conventional Cumulus Measure of Density in Breast Cancer Risk Prediction: A Literature Review.

Mammographic textures show promise as breast cancer risk predictors, distinct from mammographic density. Yet, there is a lack of comprehensive evidence to determine the relative strengths as risk predictor of textures and density and the reliability of texture-based measures. We searched the PubMed database for research published up to November 2023, which assessed breast cancer risk associations [odds ratios (OR)] with texture-based measures and percent mammographic density (PMD), and their discrimination [area under the receiver operating characteristics curve (AUC)], using same datasets. Of 11 publications, for textures, six found stronger associations (P < 0.05) with 11% to 508% increases on the log scale by study, and four found weaker associations (P < 0.05) with 14% to 100% decreases, compared with PMD. Risk associations remained significant when fitting textures and PMD together. Eleven of 17 publications found greater AUCs for textures than PMD (P < 0.05); increases were 0.04 to 0.25 by study. Discrimination from PMD and these textures jointly was significantly higher than from PMD alone (P < 0.05). Therefore, different textures could capture distinct breast cancer risk information, partially independent of mammographic density, suggesting their joint role in breast cancer risk prediction. Some textures could outperform mammographic density for predicting breast cancer risk. However, obtaining reliable texture-based measures necessitates addressing various issues. Collaboration of researchers from diverse fields could be beneficial for advancing this complex field.

High-strength alginate fibers wet-spun from pre-crosslinked sodium alginate solutions

Facing the severe problem of microplastic pollution, there is an urgent need to develop biodegradable fibers to replace the petrochemical fibers. Sodium alginate, a biomass polysaccharide, has gained widespread attentions recently for the fiber manufacture. However, the limited mechanical strength of alginate fibers restricts their usages as load-bearing fabrics and reinforcement fibers. Here, we develop a novel strategy to prepare alginate multifilaments using pre-crosslinked sodium alginate solutions. The increase in the pre-crosslinking ratio effectively hinders the disentanglement of sodium alginate chains at high stretches, causing an increase in the shear viscosity of the solution ascertained from the capillarity-driven thinning process from 4.5 Pa·s to 9.9 Pa·s and facilitating the high alignment and orientation of sodium alginate chains. The resultant fibers possess a breaking strength of 474 MPa, elongation at break of 16 %, Young's modulus of 14.4 GPa, and toughness of 51.8 MJ/m3, exceeding most biomass fibers without reinforcement additives. The high orientation degree of 0.865 and high spinnability of alginate multifilaments enable their applications in multi-channel encryption fabrics that exhibit distinct information under various optical conditions. This rheological regulation of spinning solutions provides a facile yet effective strategy to enhance the mechanical performance and broaden application scenarios of alginate fibers.

CogAware: Cognition-Aware framework for sentiment analysis with textual representations

Sentiment analysis has become an important research area in artificial intelligence. Recently, the integration of sentiment analysis with cognitive neuroscience in natural language processing (NLP) tasks has attracted widespread attention. Cognitive signals and textual signals (i.e. word embeddings) both contain distinctive information for sentiment analysis tasks. However, most previous studies cannot effectively capture the specific features and cross-domain features while integrating cognitive signals acquired from brain activity and textual signals obtained from natural language processing (NLP). To address this issue, we propose CogAware, which learns to obtain a deep representation that combines purified specific features with cross-domain features from textual and cognitive signals. CogAware employs four private encoders to extract specific or cross-domain features from textual and cognitive signals alternately. It also employs feature reinforcement and orthogonality regularization to separate specific and cross-domain features from each modality. Moreover, a shared encoder and a modality discriminator are used to further capture cross-domain features from different modalities. Our designed architecture utilizes cognitive signals and word embeddings during model training, yet relies solely on word embeddings for model inference. Experiments on a public dataset show that CogAware achieves new state-of-the-art performance on the sentiment analysis task compared with other existing models. The source code of CogAware is available at: https://github.com/zhejiangzhuque/CogAware.

REEGNet: A resource efficient EEGNet for EEG trail classification in healthcare

The architecture of REEGNet (Recurrent EEG Network) combines convolutional blocks from EEGNet with long short-term memory (LSTM) layers in order to improve temporal modeling capabilities. Additionally, the use of depthwise convolutions enables REEGNet to learn spatial filters specific to distinct frequency information. When evaluated on the exceptionally challenging IMAGENET configuration of the MindBigData Brain Dataset, consisting of 70,060 samples across 5 channels sampled at 128 Hz, REEGNet outperformed state-of-the-art models including EEGNet and LSTM architectures across all class configurations ranging from 2 to 40 classes. Specifically, REEGNet surpasses EEGNet by a margin of 10.6% in multi-class classification scenarios such as configurations with 10 classes, achieving an accuracy of 27.9%. These results highlight REEGNet’s potential to increase the accessibility of electroencephalography (EEG) analysis by allowing for reliable classification without extensive computational resources. This makes REEGNet uniquely suited for deployment in resource-constrained real-world environments and applications.

Generalized nonlinear hybrid-norm parallel sparse filtering for bearing fault diagnosis under complex interference

Abstract Considerable attention has been garnered by the bearing fault diagnostic method, founded on sparse feature learning, for its exceptional robustness and adaptability to noise. Being an effective sparse learning method, parallel sparse filtering (PSF) has found successful applications in intelligent fault diagnosis. However, influenced by mechanical structural collisions and complex transmission paths, the obtained bearing signals come with abnormal impacts and noise, resulting in a reduced reliability of the PSF. Additionally, PSF constrains features and samples with l 1 / 2 norm, which is significantly disturbed by noise. To address these issues, generalized nonlinear hybrid-norm PSF (GNHNPSF) is proposed. Firstly, the Sigmoid nonlinear function is examined to suppress the amplitude of the anomalous shock. Subsequently, the hybrid-norm constraints, which involve the application of a l 1 / 2 norm to the column features of the input signal and a l 1 / 3 norm to the row features, can be employed to effectively extract distinctive information from the original data. The GNHNPSF is demonstrated to improve the accuracy of bearing fault diagnosis under complex interference, as evidenced by the results of simulation and experiments.

A novel multi-label feature selection method based on knowledge consistency-independence index

Multi-label classification encounters the challenge of dealing with high dimensional data. In response to this challenge, numerous researchers have proposed various multi-label feature selection methods from different perspectives. However, existing methods overlook the consistency and independence of knowledge granules, and thus fail to extract valuable and distinctive information from the knowledge granules that is relevant to the label space. To address this issue, we propose a novel multi-label feature selection method based on the knowledge consistency-independence index (CIMLFS). Firstly, we introduce the concepts of knowledge consistency granularity and knowledge independence granularity to explore valuable and distinctive information from the knowledge granule families. Secondly, based upon these concepts, we define the consistency coefficient, independence coefficient, and consistency gain for features, ultimately considering the three perspectives to achieve the knowledge consistency-independence index. Furthermore, we present a multi-label feature selection method utilizing the index. Finally, to assess the effectiveness of CIMLFS, we conduct comparative experiments with eight representative multi-label feature selection methods on twelve benchmark multi-label data sets and using four evaluation metrics. The final experimental results indicate that CIMLFS ranks the first on three metrics and the second on one metric.

A Survey of Distinctive Linguistic Information for Dentists to Reach a Diagnosis as Acute Suppurative Pulpitis-The Difference in Judgements between Beginners and Experts-

A Survey of Distinctive Linguistic Information for Dentists to Reach a Diagnosis as Acute Suppurative Pulpitis-The Difference in Judgements between Beginners and Experts-

Benchmarking the transcriptomic predictive biomarkers for immunotherapy plus chemotherapy: Results from phase III RATIONALE-309 and ORIENT-11 randomized trials.

e14624 Background: The combination of PD-(L)1 inhibitors and chemotherapy (IO+Chemo) has become the standard first-line treatment for multiple types of cancer. In contrast to PD-(L)1 monotherapy, there is a lack of established biomarkers to distinguish the outcomes of IO+Chemo. The paucity of high-quality omics data in this setting has not allowed for deep biomarker exploration. We utilized RNA-seq data from two in-house randomized controlled trials (RCTs), RATIONALE-309 and ORIENT-11, with the goal to find transcriptomic biomarkers to identify patients who may not derive benefit from IO+Chemo. Methods: RATIONALE-309 (NCT03924986) was a phase 3 study conducted at 42 sites in Asia. Treatment-naive recurrent/metastatic nasopharyngeal cancer patients were randomly allocated to either PD-1 inhibitor tislelizumab or placebo plus gemcitabine and cisplatin. ORIENT-11 (NCT03607539) was a phase 3 study that enrolled patients from 47 centers in China who had previously untreated locally advanced/metastatic nonsquamous non-small cell lung cancer. Patients were randomized to PD-1 inhibitor sintilimab or placebo plus pemetrexed and platinum. We compared several immune-related biomarkers, including IFN-γ signature, PD-L1, cytolytic score (CYT), T-cell inflamed gene expression profile, B cell infiltration, and ESTIMATE immune score, and explored the optimal predictive threshold. Results: In the RATIONALE-309 and ORIENT-11 trials, 247 and 171 patients with RNA-seq data were analyzed, respectively. Although the addition of Chemo to IO could theoretically benefit tumors with low immune activation, we hypothesized that patients might not benefit from IO+Chemo if their activation level is extremely low. We first tested the predictiveness of several thresholds (20-50th quantiles) for each selected biomarker in the two RCTs. We found that low CYT and B cell are the two best indicators for identifying a lack of survival benefits from IO+Chemo versus Chemo, especially when using 20-30th quantile as threshold. To incorporate their distinct biological information and predictive value, we combined their genes into a single gene signature to represent immune activation. We found that patients in the lower quantile (&lt;20th) showed no benefit of IO+Chemo over Chemo in terms of overall survival (OS) (RATIONALE-309: HR=1.01, P=0.99; ORIENT-11: HR=1.54, P=0.31) and progression-free survival (PFS) (RATIONALE-309: HR=0.76, P=0.42; ORIENT-11: HR=1.30, P=0.52). Moreover, this lack of advantage was maintained in key subgroups such as PD-L1-positive patients and former/current smokers across the two RCTs. Conclusions: Based on the two large-scale RCTs, our findings have implications for the selection of biomarkers when identifying suitable candidates for IO+Chemo treatment. B cells and CYT are crucial components required for the efficacy of IO+Chemo.

A physics-informed deep learning model to reconstruct turbulent wake from random sparse data

This study develops a flexible deep learning framework aimed at reconstructing the global turbulent wakes from the randomly distributed sparse data. The framework is based on a Generative Adversarial Networks where the generator utilizes U-Net architecture and a constraint module is integrated into the training process. It is designed to overcome challenges posed by the chaotic behavior of turbulent fields, randomness in sensor layouts, and sparse sensor numbers. The efficacy of the model is validated across three high-fidelity datasets, including laminar wake behind a circular cylinder, turbulent wake behind a circular cylinder, and turbulent wake behind a square cylinder. The proposed model demonstrates the ability to accurately reconstruct flow patterns of both turbulent and laminar wakes, even utilizing merely 0.043% of the data from the target flow field. The proposed model exhibits significant generalization capability, which means that the model has a nearly independence from the distributions of sensors and a robust adaptation across the inputs with unseen sensor numbers. Ablation studies elucidate the distinct and complementary roles of each module within the model. Additionally, the behavior of the bottleneck tensor is analyzed through visualization, including comparisons with the lift coefficient, quantitative analyses and dimensionality reduction. These visualizations confirm the ability of the model to extract distinctive phase information reliably from sparse data, thereby guiding the reconstruction of global flow patterns. These findings highlight the potential of the model for applications in fluid dynamics where data is collected in a variable manner.