Browse Articles

Capsule-enhanced hierarchical vision transformers for rare disease classification from medical images

Abstract Automated medical image analysis plays a vital role in rare disease detection, yet existing deep learning models often struggle with severe class imbalance, limited labeled data, and subtle morphological variations. To address these challenges, this paper proposes Swin-CapsuleNet, a hybrid architecture that integrates a hierarchical Swin Transformer with capsule-based representations, tailored for rare disease classification. The framework integrates a Swin Transformer backbone for multi-scale contextual feature extraction with a capsule-based classification head that preserves part–whole spatial relationships through dynamic routing. A class-balanced capsule loss is introduced to improve sensitivity toward under-represented disease categories. Extensive experiments conducted on a multi-center rare disease dataset demonstrate that Swin-CapsuleNet consistently outperforms state-of-the-art CNN, transformer, and capsule-based baselines. The proposed model achieves 94.1% accuracy, a 93.2% F1-score, and an AUC of 0.972, while attaining a macro-F1 of 0.899 for rare disease classes. Ablation studies validate the complementary contributions of hierarchical attention, capsule representations, and the proposed loss function. Furthermore, computational analysis shows that Swin-CapsuleNet offers a favorable balance between performance and efficiency, supporting its applicability in real-world clinical decision-support systems.

V ₂ O ₅ Surface Electronic Structure Suppresses Ethane Over‐Oxidation, Enabling 65% Ethylene Yield

ABSTRACT Electrochemical oxidative dehydrogenation (ODH) of ethane in solid oxide electrolysis cells (SOECs) offers an energy‐efficient route to ethylene but faces a trade‐off between conversion and selectivity due to over‐oxidation. Conventional voltage–current regulation can suppress deep oxidation but inevitably compromises ethane conversion. Here, we engineer surface electronic structures by depositing a V 2 O 5 layer on SrFe 0.9 Ti 0.1 O 3−δ (STF), introducing intrinsic O 2p (‐1.33 eV) and V 3d (‐0.18 eV) states closer to the Fermi level than in STF (‐1.49/‐4.52 eV). Density functional theory and operando infrared spectroscopy reveal three synergistic effects: enhanced ethane adsorption (Δ E ads ‐0.33 vs. ‐0.11 eV), reduced first dehydrogenation barrier (Δ G 1 1.13 vs. 1.15 eV), and promoted ethylene desorption ((Δ G des ‐Δ G 3 ) ‐4.98 vs. ‐1.92 eV). The optimized anode delivers 65% yield and 90% selectivity at 750°C, exceeding unmodified STF by 10%. This work highlights band‐center engineering as a promising design concept for regulating hydrocarbon electrode reactions.

Circulating monocytes from aseptic loosening patients exhibit enhanced tropism toward periprosthetic tissues in NOD-SCID mouse model

Deoxygenative Functionalization of Unprotected Phenols by Linear Paired Electrolysis

ABSTRACT Phenols are abundant aromatic feedstocks widely used in organic synthesis and are commonly obtained from lignin and other renewable resources. However, the functionalization of their inert C(sp 2 )─OH bonds remains a formidable challenge, often necessitating additional preactivation steps or harsh reaction conditions. Here, we report an electrochemical strategy that enables the cleavage of phenolic C(sp 2 )─OH bonds in a formally direct manner under mild and sustainable conditions through a linear paired electrolysis, allowing both hydrodeoxygenation and, for the first time, deoxygenative alkylation with alkenes. This method is compatible with a range of functional groups and complex molecular architectures, including natural product‐derived and biomass‐derived phenols, providing a general and sustainable platform for phenol valorization without requiring additional operations to convert the substrates into their protected derivatives. Mechanistic investigations reveal that an amino phosphine with low oxidation potential and water are essential to the success of this strategy, enabling the in situ formation of a key phosphinate intermediate that subsequently undergoes reduction at the cathode, either directly or mediated by 2‐methylbenzonitrile ( M4 ) to drive the overall transformation.

NLRP3 inflammasome characterization in first-trimester, preterm, and term human villous placenta

Abstract Precise regulation of inflammatory signaling at the maternal-fetal interface is essential for a healthy pregnancy, yet how the NLRP3 inflammasome is regulated in the placenta during pregnancy and in preterm birth remains poorly understood. We characterized NLRP3 inflammasome-associated signaling in human villous placental tissue from first-trimester (8–11 weeks), preterm (28–<37 weeks), and term (38–40 weeks) pregnancies. Gene expression, protein abundance, caspase-1 processing, and tissue cytokine concentrations were assessed using qPCR, western blot, and ELISA, alongside analysis of upstream inflammatory pathways. Distinct patterns were observed at multiple molecular levels. First-trimester placentas showed the highest cytokine concentrations and NLRP3 protein abundance despite low transcript levels. Preterm placentas formed a separate transcriptional cluster and exhibited accumulation of inflammasome precursors, including pro-caspase-1 and full-length gasdermin D, without corresponding caspase-1 activation. Term placentas displayed increased transcriptional activity but reduced protein abundance and IL-1β levels. These findings demonstrate that the villous placenta exhibits distinct inflammasome profiles in first-trimester, preterm, and term pregnancies, with first-trimester placentas showing high cytokine concentrations, and both preterm and term placentas displaying a state of transcriptional priming without proportional downstream activation — consistent with a ‘primed but restrained’ inflammasome phenotype in later gestation.

Differential expression of neuronal function genes follows a tissue-specific temporal dynamic during Deformed Wing Virus infection in honey bees

Multi-sensor satellite data fusion and machine learning for Eucalyptus mapping in Meket district, Ethiopia

Vapor‐Controlled Full‐Color Luminescence in Platinum(II) Complexes Bearing <i>N</i> ‐Heterocyclic Carbene and a Flexible Alkyl Chain

ABSTRACT Platinum(II) complexes bearing aromatic ligands have garnered extensive attention as they exhibit highly efficient, stimulus‐responsive luminescence originating from the triplet metal–metal‐to‐ligand charge transfer state when stacked, which enables electronic interactions between platinum ions. However, achieving full‐color luminescence changes in a single system through external stimuli remains challenging. In this work, we synthesized a flexible platinum(II) complex bearing an N ‐heterocyclic carbene ligand with a medium‐sized alkyl chain (pentyl group), [Pt(CN) 2 (C5im‐Mepy)](C5im‐MepyH + = 3‐pentyl‐1‐(4‐methyl‐(2‐pyridinyl))‐1 H ‐imidazolium, PtC5 ), thereby developing a material that displays full‐color luminescence changes in response to various vapors. The pentyl group adopted diverse conformations within the crystal, enabling the construction of multiple stacking structures within a single complex—a feature unattainable with conventional platinum(II) complexes. Furthermore, reversible structural transformations of PtC5 , resulting in distinct luminescence color changes, were induced in the solid state through controlled vapor exposure. Detailed structural and spectroscopic studies elucidated the mechanism underlying the full‐color vapochromism—achieved for the first time in this system—, and the influence of the alkyl chain was clarified through a comparison with related complexes. This research provides valuable guidelines for developing advanced luminescent materials with precise and diversely tunable features.

Effects of acute ebselen add-on treatment on resting state function connectivity in depressed patients with inadequate response to antidepressants

Abstract Ebselen has been shown to induce a positive emotional processing bias and brain neurochemistry in healthy volunteers. However, little is known about its effects on resting-state functional connectivity (rs-FC) in depressed patients who show inadequate responses to standard antidepressants. The study aimed to investigate the effect of ebselen as an add-on treatment on rs-FC in this population. We conducted a randomised, double-blind, placebo-controlled trial involving depressed patients who were not responsive to their current treatment. Participants received either ebselen 600 mg twice daily or placebo for seven days. Resting-state functional magnetic resonance imaging was conducted at baseline and on day seven. Resting-state data were analysed using seed-based analysis and independent component analysis. Greater changes in rs-FC between the pregenual anterior cingulate cortex and the cerebellum were observed in ebselen than in placebo, but became negligible after corrections for multiple comparisons. We identified treatment effects within visual networks and between visual and sensorimotor networks, however, these did not survive correction for multiple comparisons. Short-term ebselen treatment in depressed patients with inadequate antidepressant responses did not produce significant changes in rs-FC. These findings may reflect limitations related to the short treatment duration and small and heterogeneous sample.

Graphdiyne Oxide‐Enabled Solid‐State Proton Battery Exhibiting Superior Rate Capability

ABSTRACT Proton batteries are emerging as a promising next‐generation energy storage technology, featuring high safety, superior power density, and excellent stability. However, the development of solid‐state electrolytes with high proton conductivity remains a significant challenge. Herein, graphdiyne oxide (GDYO) is proposed as a novel solid‐state electrolyte for proton batteries for the first time. By constructing a three‐dimensional hydrogen bonding network with phosphoric acid, the resulting GDYO@H 3 PO 4 electrolyte effectively addresses the low conductivity issue of conventional solid‐state electrolytes at room temperature. Fourier‐transform infrared (FTIR) spectroscopy and solid‐state fluorescence spectroscopy confirm the formation of an interconnected hydrogen bonding network between GDYO sheets and H 3 PO 4 , facilitating efficient proton transport via a hopping transport mechanism. The assembled hydrous vanadium hexacyanoferrate//MoO 3 (VHCF//MoO 3 ) full solid‐state proton battery demonstrates superior rate capability, retaining 53.6% of its capacity retention at an ultrahigh charge–discharge rate of 50C. Furthermore, the battery delivers remarkable cycling stability with a specific capacity of 102.2 mAh g − 1 and 90.2% capacity retention after 4500 cycles, significantly outperforming state‐of‐the‐art systems. This research will provide a new way to construct solid‐state proton battery with long cycling stability and high performance, and highlight the exceptional potential of GDYO as a molecular‐level design platform for advanced electrocatalysts in solid‐state batteries.

A reaction-based fluorescence for monitoring β-Gal in living cells and BALB/c nude mice

Synergistic effects of pigments, morphology, and physio-biochemical parameters reveal petal coloration differences in eight rose cultivars

A combinatorial therapeutic platform based on graphene–collagen cryogel and apoptotic vesicles for modulating hypertrophic scars

Metasurface loaded multi-port tunable silicon–graphene based THz antenna with LHCP/RHCP characteristics, low-RCS and low-coupling features

Abstract The design of a two-terminal antenna with decreased radar cross section (RCS) characteristics and polarization variety is shown in this work. By using stair-aperture feed to excite a cylindrical silicon dielectric resonator, circular polarization is accomplished. An aperture arrangement that is mirror-oriented between terminal-1 and 2 improves isolation by 30 dB. A frequency selective surface based on a split-ring resonator (SRR) is positioned between the twin ports and around the antenna structure in order to reduce monostatic RCS (MRCS). The radiator functions between 2.45 and 3.35 THz, with a 3-dB axial ratio between 2.65 and 3.15 THz. In the operational band, the MRCS is lowered by around 30 dBsm. The suggested design is well suited for radar applications in THz regime due to its broadside radiation and advantageous diversity characteristics.

Daily briefing: Human detritus remakes geology

Clinical and pathological predictors of early recurrence after neoadjuvant cisplatin gemcitabine in muscle invasive bladder cancer in the RealBLADDER study

Briefing Chat: Testosterone and sperm might get a boost from obesity drugs

GLANet: global and local anomaly network for distributed cyber threat detection using FL

Abstract This paper introduces GLANet, a novel FL-based framework designed for real-time, privacy-preserving anomaly detection in distributed network environments. GLANet is a FL-based framework that integrates a lightweight convolutional neural network (CNN) as a local anomaly detector at each distributed node with a global model aggregation mechanism based on federated averaging (FedAvg). The framework incorporates consistency regularization to align local and global model parameters, ensuring that node-specific threat patterns are preserved while enabling network-wide generalization. Differential privacy is employed through the injection of calibrated Gaussian noise into model updates before transmission to the central server, providing formal privacy guarantees without compromising detection performance. The experimental findings reveal that GLANet achieves a high detection accuracy of 97.8% on the CICIDS 2017 dataset, surpassing traditional FL baselines that achieved 92.7%. The model also achieves superior precision (96.5%), recall (97.0%), and F1 Score (96.8%), reflecting a well-balanced anomaly detection performance. Additionally, GLANet reduces communication costs by 37.5% and achieves a lower privacy loss of ε = 0.8 compared to 1.5 in baseline methods. These findings demonstrate the potential of GLANet as an effective, scalable, and secure anomaly detection framework for comprehensive defense against evolving cyber threats in dynamic and distributed network environments.

HybridViT for robust wheat leaf disease detection using CLAHE and attention-based feature fusion

EEG activities and connectivities associated with surtitle presentation during live opera viewing: an exploratory analysis of cognitive absorption