Browse Articles
Discover research articles across all indexed journals
Regenerative repair is connected to early and specific structural, immune, and metabolic MSC signatures in adult mammals
Abstract Adult mammals exhibit a limited capacity for tissue regeneration following injury and typically heal through scar formation. Mesenchymal stem/stroma cells (MSCs), which are phenotypically plastic and ubiquitous across tissues, play a critical role in maintaining tissue architecture during repair. We hypothesized that early events in regenerative and non-regenerative repair involve changes in MSC heterogeneity, which in turn determine repair outcomes. To test this hypothesis, we performed extensive single-cell RNA sequencing (scRNA-seq) in a mouse model of tissue injury. This model standardizes the comparison of non-regenerative and regenerative repair in adults with identical developmental stages and genetic backgrounds. Our analysis of MSCs during the early phases of tissue repair in adult mammals enabled the identification of distinct regenerative and non-regenerative MSC clusters, suggesting that specific MSC states may actively drive tissue repair outcomes. Furthermore, unsupervised approaches allowed us to revisit the functional signatures of MSCs centering on their impact on tissue structure (S), inflammation/immunity (I) and metabolism (M). By integrating these S, I and M functions, the SIM framework provides a conceptual model to interpret MSC behavior as a coordinated tissue-level response rather than a collection of isolated pleiotropic activities. This work positions MSCs at the center of the “SIM” triad, underscoring their pivotal role in tissue repair.
Machine learning model for predicting rebleeding risk after endoscopic variceal ligation in esophageal variceal bleeding
Abstract Rebleeding is a severe complication following recovery from esophageal variceal bleeding (EVB), yet robust predictive tools for assessing post-treatment risk after endoscopic variceal ligation (EVL) therapy remain scarce. This study developed and independently validated a machine learning (ML) model using multidimensional clinical data to predict 1-year rebleeding risk. Two independent cohorts were included: a retrospective cohort ( n = 373) for model development and a prospective cohort ( n = 119) for validation, with a one-year rebleeding endpoint. Predictors were identified using Recursive Feature Elimination (RFE), and eight ML algorithms were evaluated. Each algorithm was optimized via 5-fold cross-validation. The model with optimal performance was chosen to develop an online computational platform. RFE identified eight key predictors. The XGBoost model demonstrated superior predictive performance in both the training and validation cohorts, achieving AUCs of 0.883 and 0.887, respectively. This model was subsequently implemented in an online clinical platform for individualized 1-year rebleeding risk assessment. Our findings establish XGBoost as an effective tool for predicting EVB rebleeding risk, providing an evidence-based decision aid for post-EVL management.
MSCA-TNet based deep learning method for ECG arrhythmia classification
An Attention-Enhanced ViT-HLNN Hybrid Ensemble Framework for Multi-Class Gastrointestinal Disease Classification
Design and field validation of an automated sensor-integrated high-capacity corn plot seeder
Secure task offloading framework for industrial edge computing using reconfigurable intelligent surfaces and spectrum agility
UV-C irradiation enhances the dual-use potential of Desmodesmus quadricauda and Scenedesmus dimorphus for food and biodiesel production
Biomimetic macrophage membrane-coated nanoparticles for targeted cardioprotection against myocardial ischaemia-reperfusion injury
Clinical registry metadata as a hidden bottleneck in AI-driven drug discovery: a computational audit of translational phase data in glioma research
Development and properties of composite coatings for surface repair of cement concrete pavements
Design of three-mode free-form nanostructured optical fibers: comparison of dense and convolutional neural networks in Generative Inverse Design Networks approach
Abstract We report a numerical study on the inverse design of the internal structure of weakly coupled three-mode fibers. We explored a new class of optical fibers – free-form nanostructured fibers (FFNFs) – operating at 1550 nm for potential application in Mode-Division Multiplexing (MDM) systems. The fiber geometries were generated and optimized within the Generative Inverse Design Networks (GIDNs) framework using convolutional neural networks (CNNs) and fully connected dense neural networks (DNNs). The objective of the optimization was to maximize the minimal effective refractive index separation Min|Δ n eff | between supported modes, ensuring weak intermodal coupling. The proposed free-form nanostructured fiber designed with the CNN achieved a minimum modes separation of Min|Δ n eff | = 2.15 × 10 − 3 , exceeding that of a reference three-mode elliptical-core fiber (Min|Δ n eff | = 2.085 × 10 − 3 ). In contrast, the best DNN-optimized structure reached Min|Δ n eff | = 1.99 × 10 − 3 . The results demonstrate that the CNN-based inverse design yields fiber geometries outperforming conventional designs. The proposed methodology can be extended to higher-order mode systems and can include additional fiber properties crucial for telecommunication purposes.
Research on intelligent generation of volleyball training strategies combining YOLOv5 + DeepSORT trajectory data and a key point CNN model
Machine learning for scalable obstructive sleep apnea risk screening using digital phenotyping from wearable devices and clinical scales
Interacting dust-acoustic multi-soliton and periodic waves in strongly coupled dusty plasma with ion-drag force
Identification and characterization of intracerebral hemorrhage events in elderly veterans with alzheimer’s disease in the veterans affairs healthcare system
Abstract Cerebral amyloid angiopathy (CAA) and hypertensive (HTN) small vessel disease are causes of spontaneous intracerebral hemorrhage (ICH). We identified ICH rates in patients with all-cause mild cognitive impairment (MCI) or Alzheimer’s disease (AD), and explored feasibility of using location-based approach to differentiate CAA- and HTN-related ICH using comorbidities in electronic health records (EHRs). This administrative study combined Veterans Affairs Healthcare System plus Centers for Medicare and Medicaid Services (VAHS/CMS) databases. Patients with MCI/AD aged ≥ 50 years (2016–2023) were 1:1 matched to non-MCI/AD controls. Inpatient primary discharge International Classification of Diseases-10th Edition (ICD-10) codes identified acute ICH; anatomical location within codes classified events as likely CAA- or HTN-related ICH. Incidence rates of ICH after MCI/AD were summarized. Cluster analysis of variables related to ICH was used to describe whether data-driven groupings matched clinician-postulated classifications. The MCI/AD cohort ( n = 747,475) and controls were aged 77.7 ± 10.1 years (96% men, 75–76% White, 87–88% non-Hispanic). Demographic- and comorbidity-adjusted rates of ICH/1000 person-years were 0.84 (overall), 1.05 (MCI/AD), and 0.68 (non-MCI/AD). Adjusted events/1000 person-years were higher in MCI/AD vs. non-MCI/AD cohorts: CAA-related ICH, 0.19 vs. 0.12 (Incidence Rate Ratio (IRR) 1.63; P < 0.001); HTN-related ICH, 0.16 vs. 0.12 (IRR 1.35; P < 0.001); non-specific-ICH, 0.68 vs. 0.43 (IRR 1.59; P < 0.001). Hierarchical clustering analysis of our cohorts revealed an association of CAA-related ICH with older age, cardiovascular and rheumatic disorders, and an association of HTN-related ICH with cerebrovascular disease, hypertension and diabetes. In sum, the estimated incidence of ICH over the study period was 0.84/1000 person-years. CAA-related ICH incidence in MCI/AD was 63% greater than that in controls. Outcomes from cluster analysis are consistent with ICD-based CAA- vs. HTN-related ICH classifications. These findings support future exploration of using ICD coding-based ICH event identification in EHRs and claims databases for epidemiological studies.
Confidence-guided outlier refinement and collaborative embedding for unsupervised person re-identification
Abstract Unsupervised person re-identification techniques have developed rapidly in recent years. Nonetheless, they still face challenges such as unstable pseudo-label quality and insufficient feature representation, particularly when handling outlier points and complex backgrounds. To address these issues, this paper proposes a joint optimization algorithm of Multi-level Confidence Outlier Refinement (MLCOR) and Collaborative Embedding Method (CEM), which aims to improve the discriminative nature of the embedding space and optimize the accuracy of pseudo-labels. Specifically, Multi-level Confidence Outlier Refinement evaluates the confidence level of outlier points by analyzing the distance relationship between outlier points and their neighboring samples, and classifies them into multiple confidence levels. We design a weighted voting strategy for low-confidence samples to correct the pseudo-labeling of low-confidence points by using the label distribution of neighboring samples, thus reducing noise interference and clustering errors and improving the accuracy of pseudo-labeling. Meanwhile, the Collaborative Embedding Method jointly optimizes global and local features, establishing an effective synergy between global category differentiation and local fine-grained feature learning. By integrating multi-level similarity relationships, this approach not only strengthens the model’s ability to capture subtle differences between samples but also significantly enhances the model’s boundary awareness. Experimental results demonstrate that the proposed method achieves outstanding performance on multiple standard datasets, significantly improving both clustering accuracy and pseudo-label precision, while also exhibiting strong domain generalization and robustness in complex environments.