Modulating textures jointly in orientation and spatial frequency makes them easily distinguishable from the surround. The performance benefit of double-cue targets in detection and discrimination tasks is stronger than expected from independent feature processing, known as "feature synergy". To explore the neural origin of this effect, we had 38 observers perform a texture figure localization task and a more demanding shape identification task, while simultaneously recording EEG. The results showed a strong feature synergy effect in both tasks, which was accompanied by significantly reduced posterior ERP amplitudes in a cluster of 13 adjacent electrodes from the left, central and right occipital and central parieto-occipital lobes. The double-cue specific amplitude reduction occurred within a time window ranging from 200 to 290 ms around the P2 (TOI-1) and, to a lesser extent, at later times ranging from 290 to 380 ms, including the P3 peak (TOI-2). In TOI-1, but not in TOI-2, the cluster electrodes responded to enhanced figure-ground segregation and also encoded the perceptual summation of this effect for double-cue targets. Moreover, ERP reduction was stronger for localization than for shape identification in TOI-1, but the effect was reversed in TOI-2, where significant double-cue effects mainly concerned shape identification. Different task influences on the EEG correlate of feature synergy during earlier and later time periods indicate that fewer resources are necessary for a given task when targets are redundantly defined. This suggests an origin in sites where features and shapes are processed under attentional control.

Apathy is a common and debilitating neuropsychiatric symptom in Huntington's disease (HD), yet its long-term trajectory remains poorly characterized. This study examined changes in apathy in HD gene expansion carriers (HDGECs) over 6years, using a multidimensional measure, and investigated associations with cognition, motor symptoms and depression. Eighty-two HDGECs (premanifest and manifest) completed assessments at Time 0 and Time 1 with a mean follow-up interval of 6years. Apathy was measured using the Lille Apathy Rating Scale (LARS) and the Problem Behaviors Assessment-short (PBA-s). Depressive symptoms were assessed with the Hamilton Depression Rating Scale, while social cognition and executive functioning were measured using the Emotion Hexagon and Symbol Digit Modalities Test. Within-person changes were examined using paired statistical tests and associations with clinical variables were evaluated using correlation analyses. Total apathy scores increased significantly over 6years, with small changes (LARS: 1.38 points; PBA-s: 1.11 points). Premanifest participants showed a selective decline in Action Initiation, whereas manifest participants exhibited a broader worsening of total apathy. Individual trajectories were variable, with both worsening and improvement observed. An increase in depressive symptoms was significantly correlated with changes in apathy but accounted for little variance. Apathy in HD shows a small but significant increase over 6years, characterized by marked heterogeneity and changes in Action Initiation. The limited association with depression and cognition highlights apathy as an independent neuropsychiatric feature. These findings underscore the heterogeneous nature of apathy progression and the value of multidimensional assessment in longitudinal studies.

Differentiation of Focal Nodular Hyperplasia From Hepatocellular Adenoma on Contrast-Enhanced US With Perfluorobutane or Sulfur Hexafluoride.

Clinical and MRI features for differentiating reactive lymphoid hyperplasia from hepatocellular carcinoma in non-cirrhotic chronic HBV patients.

Maintenance of bound or independent features in visual working memory is task-dependent.

Differentiating large-duct pancreatic ductal adenocarcinoma from malignant intraductal papillary mucinous neoplasm: MRI characteristics and diagnostic implications.

Current machine learning methods only utilize the three-channel color features of optical images for computer visual tasks. However, the optical images only explicitly present information of RGB color and two-dimensional planar shape, where the third-dimensional spatial features are not fully exploited. This limitation restricts the potential improvement in recognition performance. To address this issue, we propose a detection scheme to enhance model's detection capabilities based on four independent features by combining the pseudo-depth and the RGB features without adding any additional hardware sensors. The monocular depth estimation model is first used as a virtual depth sensor to extract the pseudo-depth features from input optical images. Then the fused Depth-RGB features are fed into the neural network model for object detection training and inference to enhance capability for extracting spatial features. Experiments show that the proposed method has improved the detection metric mAP[Formula: see text] by 3.8 and 8.0 percentage points on the public M[Formula: see text]FD and COCO datasets, respectively. Notably, the scheme can be easily embedded into any machine learning models to definitely improve the detection performance.

Metabolic dysfunction-associated steatotic liver disease (MASLD) is often asymptomatic early on but can progress to irreversible conditions like cirrhosis. Due to limited access to human liver biopsies, systematic and integrative molecular resources remain scarce. In this study, we performed transcriptomic analyses on liver and metabolomic analyses on liver and plasma samples from morbidly obese individuals without liver pathology or at early-stage MASLD. While the plasma metabolomic profile did not fully mirror liver histological features, dual-omics integration of liver samples revealed significantly remodeled lipid and amino acid metabolism pathways. Integrative network analysis uncoupled metabolic remodeling and gene expression as independent features of hepatic steatosis and fibrosis progression, respectively. Notably, GTPases and their regulators emerged as a novel class of genes linked to early liver fibrosis. This study offers a detailed molecular landscape of early MASLD in obesity and highlights potential targets of obesity-linked liver fibrosis.

Metabolic dysfunction-associated steatotic liver disease (MASLD) is often asymptomatic early on but can progress to irreversible conditions like cirrhosis. Due to limited access to human liver biopsies, systematic and integrative molecular resources remain scarce. In this study, we performed transcriptomic analyses on liver and metabolomic analyses on liver and plasma samples from morbidly obese individuals without liver pathology or at early-stage MASLD. While the plasma metabolomic profile did not fully mirror liver histological features, dual-omics integration of liver samples revealed significantly remodeled lipid and amino acid metabolism pathways. Integrative network analysis uncoupled metabolic remodeling and gene expression as independent features of hepatic steatosis and fibrosis progression, respectively. Notably, GTPases and their regulators emerged as a novel class of genes linked to early liver fibrosis. This study offers a detailed molecular landscape of early MASLD in obesity and highlights potential targets of obesity-linked liver fibrosis.

Metabolic dysfunction-associated steatotic liver disease (MASLD) is often asymptomatic early on but can progress to irreversible conditions like cirrhosis. Due to limited access to human liver biopsies, systematic and integrative molecular resources remain scarce. In this study, we performed transcriptomic analyses on liver and metabolomic analyses on liver and plasma samples from morbidly obese individuals without liver pathology or at early-stage MASLD. While the plasma metabolomic profile did not fully mirror liver histological features, dual-omics integration of liver samples revealed significantly remodeled lipid and amino acid metabolism pathways. Integrative network analysis uncoupled metabolic remodeling and gene expression as independent features of hepatic steatosis and fibrosis progression, respectively. Notably, GTPases and their regulators emerged as a novel class of genes linked to early liver fibrosis. This study offers a detailed molecular landscape of early MASLD in obesity and highlights potential targets of obesity-linked liver fibrosis.

Lateral patellar compression syndrome (LPCS) is characterized by abnormal pressure elevation in the lateral patellofemoral joint. The histopathological basis of LPCS, particularly regarding microvascular and inflammatory changes in the lateral joint capsule and lateral retinaculum, remains unclear. This study aimed to systematically compare the histological characteristics of these tissues between patients with LPCS and controls to characterize tissue-level features associated with LPCS and to explore the potential implications of these findings for understanding pathophysiology and informing future research on surgical strategy selection. This retrospective case–control study included 26 patients with LPCS and 23 control patients (with meniscal injuries, patellar dislocation, patellofemoral osteoarthritis, or synovial cyst). Intraoperative tissue specimens of the lateral joint capsule and lateral retinaculum were obtained. Histological evaluation was performed using immunohistochemical staining for Factor Ⅷ (F8, microvessels), CD3 (T cells), CD20 (B cells), and CD68 (macrophages). Microvessel density and inflammatory cell density were quantified and compared between groups. Microvessel density was significantly higher in both the lateral joint capsule (48.84 ± 14.12/mm2 vs. 34.72 ± 15.40/mm2, P = 0.002) and the lateral retinaculum (27.50 ± 11.33/mm2 vs. 17.13 ± 9.26/mm2, P = 0.007) of patients with LPCS compared to controls, with large effect sizes (Cohen's d > 1.0). Within both groups, microvessel density was significantly higher in the joint capsule than in the retinaculum (both P < 0.001). In contrast, no statistically significant between-group differences were detected in the densities of CD3 + T cells, CD20 + B cells, or CD68 + macrophages in either tissue. However, in multivariate regression models adjusting for age and sex, the between-group differences in microvessel density were attenuated and no longer statistically significant; interaction terms were also not significant. Unadjusted analyses showed increased microvessel density in the lateral joint capsule and lateral retinaculum of patients with LPCS while no statistically significant between-group differences were detected in classical inflammatory cell densities (CD3 + , CD20 + , CD68 +) under the current immunohistochemical panel. Because the association with microvessel density was not statistically significant after adjustment for age and sex, these findings should be interpreted as hypothesis-generating and potentially influenced by demographic factors. Larger age- and sex-matched studies, ideally with clinical correlation, are needed to determine whether angiogenesis is an independent pathological feature of LPCS and whether it relates to outcomes.

Transparent models, which provide inherently interpretable predictions, are receiving significant attention in high-stakes domains. However, despite much real-world data being collected as time series, there is a lack of studies on transparent time series models. To address this gap, we propose a novel transparent neural network model for time series called Generalized Additive Time Series Model (GATSM). GATSM consists of two parts: 1) independent feature networks to learn feature representations, and 2) a transparent temporal module to learn temporal patterns across different time steps using the feature representations. This structure allows GATSM to effectively capture temporal patterns and handle varying-length time series while preserving transparency. Empirical experiments show that GATSM significantly outperforms existing generalized additive models and achieves comparable performance to black-box time series models, such as recurrent neural networks and Transformer. In addition, we demonstrate that GATSM finds interesting patterns in time series.

To investigate subtle cognitive dysfunction in patients with cervical dystonia (CD) as a potential independent non-motor feature of the disease or as a consequence of interactions between motor and other non-motor symptoms (NMS). Cognitive impairment represents one of the most common non-motor symptoms in patients with cervical dystonia. However, the interrelations between cognitive dysfunction, motor symptoms, and other non-motor symptoms remain insufficiently explored. Patients with CD (n = 34) underwent comprehensive assessment at baseline and 4-6 weeks after botulinum toxin (BoNT) treatment. Clinical and sociodemographic variables, as well as motor and non-motor symptoms of dystonia were assessed. Matched controls (n = 33) underwent a single assessment. Cognitive function, depressive and anxiety symptoms, and sleep disturbances were assessed by a neuropsychologist in both groups. At baseline, 52.9% of patients with CD scored below the Montreal Cognitive Assessment (MoCA) cut-off for cognitive impairment, compared with 3.0% of healthy controls. Patients also showed higher rates of depressive symptoms, anxiety, and sleep disturbances. Cognitive performance was significantly lower in the CD group across multiple domains [executive functions, visuospatial abilities, language, memory, and attention (p < 0.05)]. Following botulinum toxin treatment, significant improvements were observed in overall MoCA scores and in specific domains of executive function, visuospatial abilities, language, and memory (p < 0.05). No correlation was found between overall dystonia severity and cognitive performance, although executive function correlated with motor symptom severity at baseline (R = -0.41, p = 0.017). In the multivariate model, sleep disturbances were identified as the strongest negative predictor of cognitive function (β = -0.40, p = 0.006), while higher education showed a protective effect. Other variables, including depression, anxiety, age, disease duration, and dystonia severity, were not significant predictors. The study explores cognitive impairment in cervical dystonia in relation to motor severity and non-motor domains, with the multivariable model suggesting a potential role of sleep disturbances. The findings do not allow a definitive distinction between cognitive impairment as a core feature of dystonia or a secondary effect related to non-motor symptoms, but they indicate complex and interdependent mechanisms.

Lung cancer patients are frail because they are usually diagnosed at an advanced age, are often accompanied by comorbidities, and are at high risk of chemotherapy toxicity. Choosing a safe and correct treatment is very important. In this study, we investigated the ability of VES-13, which measures patient frailty, to predict the side effects of chemotherapeutic drugs. This was a prospective, single-center observational study in which 131 patients aged ≥ 65 years who were scheduled to receive chemotherapy due to a lung cancer diagnosis were included. The VES-13 questionnaire was completed by all patients before treatment. Treatment-related toxicities (TRTs) while receiving chemotherapy treatment were assessed according to the National Cancer Institute Common Terminology Criteria for Adverse Events v4.03. The median age was 70 (65-85) years. There was hematologic toxicity in 46 (35.1%) patients and nonhematologic grade 3-5 toxicity in 33 (25.2%) patients. In the univariate analysis, VES-13 (OR = 8.40, p < 0.001), platinum chemotherapy (OR = 2.53, p = 0.016) and ECOG-PS (OR = 2.60, p = 0.018) were found to be predictive of TRT. In the multivariate analysis, VES-13 remained an independent predictive feature when the predictors were evaluated together (OR = 8.26, p < 0.001). The detection of abnormal scores with the VES-13 tool revealed a greater need for initial treatment dose reduction, treatment interruption, treatment discontinuation, blood transfusion and unexpected hospital admission (p = 0.012, p < 0.001, p < 0.001, p = 0.022, and p < 0.001, respectively). We believe that the VES-13 questionnaire is a useful tool for deciding on a chemotherapy and choosing the appropriate regimen and that it should be used in outpatient clinic conditions because it is easy and short to apply.

Deciphering visual stimuli from fMRI data presents a significant challenge in computational neuroscience. This paper introduces a novel, optimized ensemble learning framework for high-accuracy visual object recognition. Our method employs a mutual information-based hierarchical clustering technique to automatically segment the high-dimensional voxel space into independent feature facets. An ensemble of Support Vector Machine (SVM) classifiers is then trained on these facets. Crucially, the entire framework—including the number of facets, the fusion operator, and SVM parameters (C, gamma)—is globally optimized using the Simulated Annealing algorithm to ensure peak performance. We rigorously evaluated our approach on three public fMRI datasets: DS105 (8 visual objects), DS107 (4 semantic categories), and DS116 (2 visual oddball stimuli). The proposed model demonstrated exceptional performance, achieving mean recognition accuracies above 95% across all three datasets, with peak subject-level accuracy reaching 100%. Specifically, our Ensemble-965 model (using the detailed Talairach Atlas) attained accuracies of 98.6% on DS105, 97.5% on DS107, and 99.4% on DS116, surpassing current state-of-the-art brain decoding methods under comparable validation conditions. These results indicate that our method provides a robust, accurate, and highly effective solution for visual brain decoding.

ABSTRACT Multi‐step ahead forecasting of power load is crucial for optimising power system scheduling and participating in energy market transactions. However, existing forecasting methods often suffer from issues such as cumulative prediction errors and insufficient modelling of sequence dependencies. To solve the above problems, a multi‐step ahead forecasting method based on multiplexed convolutional neural networks (MCNN) and multi‐gate mixture of long short‐term memory networks (MMoL). First, multi‐branch convolution is adopted to construct independent feature spaces for different levels of power loads, achieving multi‐scale feature fusion to enhance the representation ability of the input samples. Next, the multi‐step prediction task is transformed into a multi‐task joint optimisation problem. Multiple independent LSTMs are used as shared experts, and task‐specific gating units are utilised to dynamically learn the optimal combination of expert models for each future time step, achieving more refined time‐series feature modelling. Finally, comparative experiments are conducted based on two real‐world datasets. The results show that the proposed model exhibits better accuracy and robustness.

Cognitive impairment is a prevalent condition among middle-aged and older adults and often progresses to dementia, posing substantial clinical and societal burdens. Early assessment of high-risk individuals is essential for timely intervention and management. This study aimed to develop a practical nomogram for the assessment of cognitive impairment in community-dwelling elderly populations. This cross-sectional study recruited 581 participants between October 23 and November 8, 2023, comprising 465 assigned to the training cohort and 116 to the validation cohort. Demographic information, medical history, lifestyle, and biochemical parameters were collected using structured questionnaires. Cognitive impairment was assessed via the Montreal Cognitive Assessment (MoCA). Independent features were identified using LASSO regression followed by binary logistic regression, and a nomogram was constructed based on these variables. Model performance was evaluated by discrimination, calibration, and clinical utility using Receiver Operating Characteristic (ROC) curves, calibration plots, the Hosmer–Lemeshow test, and Decision Curve Analysis (DCA). Cognitive impairment prevalence was 38.5% in the training and 32.8% in the validation cohort. Six features—sex, age, systolic blood pressure, homocysteine, fruit consumption, and family history of stroke—were integrated into the nomogram. The model demonstrated good discrimination (AUC 0.816 in training cohort; 0.796 in validation cohort) with satisfactory calibration and clinical applicability. The proposed nomogram provides a reliable and convenient tool for the early risk assessment of cognitive impairment in middle-aged and older adults, facilitating targeted prevention and personalized management in clinical and community settings. Its implementation may assist healthcare professionals in identifying high-risk individuals and mitigating progression toward dementia.

Electroencephalogram (EEG) signals serve as a primary input for brain-computer interface (BCI) systems, and extensive research has been conducted on EEG-based emotion recognition. However, because EEG signals are inherently contaminated with various types of noise, the performance of emotion recognition is often degraded. Furthermore, the use of a Band Feature Extraction Neural Network (BFE-Net), a state-of-the-art (SOTA) method in this field, has limitations with respect to independent band-wise feature extraction and a simplistic band aggregation process to obtain final classification results. To address these problems, this study proposes the noise-robust band-attention BFE-Net framework, aiming to improve the conventional BFE-Net from two perspectives. First, we implement multiple-input, multiple-output (MIMO)-based preprocessing. Specifically, we utilize multichannel minima-controlled recursive averaging for precise non-stationary noise covariance estimation and generalized eigenvalue decomposition for subspace filtering to enhance the signal-to-noise ratio. Second, we propose an attention-based band aggregation mechanism. By integrating a band-wise self-attention mechanism, the model learns dynamic inter-band dependencies for more sophisticated feature fusion for classification. Experimental results on the SEED and SEED-IV datasets under a subject-independent protocol show that our model outperforms the SOTA BFE-Net by 3.27% and 3.34%, respectively. This confirms that rigorous MIMO noise reduction, combined with frequency-centric attention, significantly enhances the reliability and generalization of BCI systems.

Cancer research emphasises early detection, yet quantitative methods for normal tissue analysis remain limited. Digitised haematoxylin and eosin (H&E)-stained slides enable computational histopathology, but artificial intelligence (AI)-based analysis of normal breast tissue (NBT) in whole slide images (WSIs) remains scarce. We curated 70 WSIs of NBTs from multiple sources and cohorts with pathologist-guided manual annotations of epithelium, stroma, and adipocytes (https://github.com/cancerbioinformatics/OASIS). We developed robust convolutional neural network (CNN)-based, patch-level classification models, named NBT-Classifiers, to tessellate and classify NBTs at different scales. Across three external cohorts, NBT-Classifiers trained on 128 × 128 µm and 256 × 256 µm patches achieved AUCs of 0.98–1.00. The model learned independent normal features different from those of precancerous and cancerous epithelium, which were further visualised using two explainable AI techniques. When integrated into an end-to-end preprocessing pipeline, NBT-Classifiers facilitate efficient downstream analysis within peri-lobular regions. NBT-Classifiers provide robust compartment-specific analytical tools and enhance our understanding of NBT appearances, which serve as valuable reference points for identifying premalignant changes and guiding early breast cancer prevention strategies.

Abstract Learning minimal interpretable models (e.g., decision trees, decision sets, and binary decision diagrams) is computationally challenging, yet increasingly important in high-stakes settings. We use decision trees as a canonical case study, but the proposed structural parameter is solver-agnostic. Recent parameterized-complexity results show fixed-parameter tractability when parameterized by model size s and a data-dependent conflict parameter $$\delta $$ δ , the maximum Hamming disagreement between oppositely labeled examples. We show that $$\delta $$ δ is highly noise-sensitive: under small relevant support and independent irrelevant features, $$\delta $$ δ typically scales with ambient dimension, making $$\delta $$ δ -based branching uninformative. We introduce a distribution-aware alternative, the effective conflict width $$\kappa _\tau $$ κ τ , obtained by restricting conflicts to features whose relevance exceeds a threshold. We instantiate this idea as structure-guided branching (SGB), which branches on relevance-filtered conflict features and safely falls back to full $$\delta $$ δ -branching. Using conflict-driven branching simulations to isolate search-tree effects, we find that $$\kappa _\tau $$ κ τ can remain stable as dimension grows and yields substantial reductions in explored search nodes on synthetic data and multiple real datasets. These results suggest structural parameters can improve the noise robustness of exact interpretable learning and can serve as solver-agnostic pruning signals.