Post-stroke rehabilitation increasingly requires technologies capable of supporting intensive, measurable, and adaptable interventions while preserving clinical supervision and patient safety. This study presents a lightweight, safety-aware, and modular multimodal robotic pipeline for embodied human-robot interaction in post-stroke rehabilitation, with a first target application in hemispatial inattention.

Computational peptide-target interaction (PTI) prediction has advanced rapidly as a tool for drug discovery, yet the field lacks agreed evaluation standards. Discrepancies in metrics, negative sampling strategies, and dataset construction across publications currently render cross-study comparisons methodologically intractable, ultimately impeding the clinical translation of therapeutic peptides. To resolve this, we present PTI-TAPE (Peptide-Target Interaction-Tasks Assessing Peptide Engagement), a standardised benchmarking framework developed via a structured three-round eDelphi consensus study with 15 international experts spanning computational biology, machine learning, and regulatory science. Operating under a robust consensus threshold (>80% agreement), the panel established 26 definitive standards across eight domains, governed by the new STRIDE (Standardisation, Transparency, Representativeness, Integration, Discovery, Evidence) framework. Key outcomes include mandating specific primary metrics for affinity and structure, enforcing a tiered negative sampling hierarchy that prioritizes experimental non-binders, and requiring mandatory temporal dataset splits. By providing a unified reporting checklist, benchmark specifications, and a biennial governance structure, PTI-TAPE ensures reproducible evaluation and fair method comparison, bridging the gap between computational PTI predictions and genuine therapeutic benefit.

Manual segmentation and evaluation of optical coherence tomography (OCT) images are difficult and rely on specialist training, which limits the output in a busy clinic. While automatic segmentation can ease this burden, the deep models that currently lead the field tend to be too heavy for routine deployment on standard clinical hardware. A further obstacle is the unusual geometry of OCT data: instead of compact objects on a background, the structures of interest form long, undulating tissue strata that extend horizontally across the whole frame, and this layout is poorly matched to convolutional kernels designed for ordinary photographs.

We synthesised a series of N2,N4-diaminoquinazoline derivatives, among which 3-[[4-(benzylamino)quinazolin-2-yl]amino]benzenesulfonamide (compound 5) exhibited potent phosphodiesterase type 5 (PDE5) inhibitory activity. Increasing evidence suggests that PDE5 inhibition is a promising therapeutic strategy for peripheral vascular disease. This study aimed to investigate vasorelaxant effects of 5 and its underlying mechanisms in rat mesenteric arteries. Compound 5 was synthesised and purified, and evaluated its vascular activity using organ bath assay. Molecular docking was performed to identify its potential molecular targets. Compound 5 (0.0001-30 μM) induced relaxation of phenylephrine (PE)-precontracted mesenteric arteries (EC50 = 0.76 ± 0.29 μM; Emax = 95.6% ± 1.1%). This effect was attenuated by endothelium removal (EC50 = 5.24 ± 0.98 μM) and by L-NAME (eNOS inhibitor). Compound 5 enhanced sodium nitroprusside-induced relaxation, supporting modulation of the NO-sGC-cGMP axis via PDE5 inhibition. Iberiotoxin (BKCa blocker) reduced relaxation, whereas 4-aminopyridine (KV blocker) and glibenclamide (KATP blocker) had no effect. Compound 5 inhibited vasoconstriction mediated by extracellular Ca2+ influx via receptor-operated Ca2+ channels (ROCCs) and α1-adrenergic receptor activation. Molecular docking provided qualitative support for PDE5A, BKCa channels and α1-adrenergic receptor as potential targets. In conclusion, 5 induces potent vasorelaxation in resistance arteries through eNOS/NO/cGMP signalling, BKCa channel activation and inhibition of ROCCs and α1-adrenergic receptors, supporting its potential as a therapeutic candidate for peripheral vascular disorders.

High-pressure drug-coated balloon (HP@DCB) angioplasty enhances vascular drug delivery by overcoming the limitations of passive diffusion-based conventional noncompliant DCB (NC@DCB) systems. We developed a dual-layer HP balloon catheter composed of an outer Nylon 12 layer and an inner Pebax 7233 layer, and evaluated its feasibility and therapeutic efficacy as a paclitaxel (PTX)-coated platform. Shellac-based PTX coating was uniformly applied to HP@DCB and NC@DCB, and coating thickness, drug amount, release behavior, and mechanical properties were compared. Pressure-assisted drug penetration was assessed in a vascular-mimicking agarose gel phantom using fluorescent PTX and confocal microscopy. In vivo efficacy was evaluated in a balloon injury-induced rabbit iliac artery stenosis model using angiography, optical coherence tomography (OCT), histology, immunohistochemistry, and immunofluorescence examinations. Drug amount, coating thickness, and sustained drug release did not differ between groups. The mechanical robustness in HP@DCB was significantly greater than that in NC@DCB. HP@DCB produced significantly greater penetration depth and fluorescence-positive area, indicating enhanced pressure-assisted drug transfer. HP@DCB resulted in superior luminal patency, reduced neointimal hyperplasia, and favorable biological responses compared with that in NC@DCB and the sham control. These findings support DCB use based on a dual-layered HP balloon platform as a promising next-generation strategy for stenotic vascular lesion treatment.

Systemic nucleic acid delivery to hematopoietic stem and progenitor cells (HSPCs) remains a major barrier to in vivo gene therapy due to hepatic sequestration of lipid nanoparticles (LNPs). Here, we report a scalable, ligand-free HSPC-tropic LNP enabling efficient mRNA delivery to human HSPCs in vivo without conditioning. In humanized NBSGW mice, the LNP preferentially delivered mRNA to human HSPCs and early myeloid progenitors, with limited delivery to lymphoid and non-hematopoietic cells. Single-cell transcriptomic and multiomic analyses revealed no broad transcriptional perturbation associated with delivery. Compared with the SM-102 LNP, the HSPC-tropic LNP achieved greater mRNA delivery to human HSPCs, while exhibiting substantially reduced hepatic delivery. The LNP also enabled functional delivery of Cas9 mRNA and β2M-targeting sgRNA to human CD34+ HSPCs, resulting in efficient gene editing and reduced β2M expression. These findings establish a ligand-free platform for functional RNA delivery and genome engineering in human HSPCs.

Reliable identification and early prediction of epileptic seizures play a critical role in improving patient outcomes and supporting timely therapeutic decision-making. To address the challenges of spatiotemporal feature coupling and long-range dependency modeling in multi-channel EEG signals, this paper proposes a channel-adaptive temporal modeling framework named GCNN-Informer. The framework follows a three-stage pipeline: a grouped convolutional network (GCNN) extracts local features and selects the five most discriminative channels; a convolutional block attention module (CBAM) performs dual spatiotemporal enhancement; and a lightweight Informer encoder captures global temporal dependencies via the ProbSparse self-attention mechanism. Experiments on the CHB-MIT dataset, conducted under a patient-specific evaluation scheme, demonstrated that the proposed method achieved accuracies of 97.33% for seizure detection and 97.28% for seizure prediction, with a low false detection rate (FDR) of 0.30/h and a false prediction rate (FPR) of 0.03/h. Grad-CAM visualization further shows that as EEG signals evolve from normal/interictal to preictal and then to ictal stages, the model's attention heatmap progressively transitions from locally sparse highlighting to continuous block-like highlighting, and finally to dense deep-red responses - a pattern that closely aligns with clinical electrophysiological dynamics. This study provides an accurate and generalizable deep-learning solution for real-time epileptic monitoring and early-warning systems.

In order to improve patient outcomes and treatment processes, early and precise diagnosis of brain tumors is crucial. Despite the strong performance shown by deep learning models in classifying brain tumors from MRI scans, integrating these models into user-friendly clinical workflows remains challenging. This study presents a web-based AI platform designed for brain tumor classification using MRI images. The platform enables users to upload MRI scans, apply standardized preprocessing, select between different deep learning models, and receive real-time predictions. A custom convolutional neural network (CNN), pretrained VGG19, and pretrained Xception were evaluated using a publicly available T1-weighted contrast-enhanced brain MRI dataset of approximately 7,000 images across four classes: glioma, meningioma, pituitary tumor, and no tumor. The dataset lacks patient demographic and clinical metadata, limiting assessment of clinical generalizability. The models achieved test accuracies of 96.03%, 98%, and 96%, respectively, with VGG19 demonstrating the best performance. The study is limited by the use of a single public, slice-based dataset and the absence of external clinical validation; future work will focus on multi-sequence MRI and independent clinical cohorts.