The natural antifungal peptide Histatin 5 (Hst 5) is a histidine-rich cationic peptide secreted by human salivary glands and a key component of oral innate immunity, but its moderate activity limits clinical use. Hst 5 enters Candida albicans via the membrane receptor Ssa1/2. Here, we integrated artificial intelligence-assisted and computer-aided drug design to rationally modified the sequence structure of Hst 5. Truncated derivatives of Hst5 were screened for antimicrobial potential using ESM2-AFPpred, and high-probability candidates were docked with Ssa1/2. The Hst 5-22 was identified, then redesigned based on alanine scanning to yield the optimized derivative Hst 5-22-RW. Compared with Hst 5, Hst 5-22-RW has a shorter sequence, stronger Ssa1/2 binding, and improved activity against C. albicans. It also shows superior activity against fluconazole-resistant strains. RT-qPCR and transmembrane tracking confirmed higher cellular transport efficiency in C. albicans. The CADD/AIDD-driven optimization successfully generated the highly active antifungal peptide Hst 5-22-RW, providing a novel strategy for rational modification of antimicrobial peptides.
With the rapid integration of AI into higher education, teachers' psychological responses are critical for technology adoption. This study examines AI self-efficacy and AI anxiety among university teachers in a Chinese university. It investigates the relationship between these two constructs and explores differences based on gender, age and academic major. A quantitative survey was administered to 350 teachers selected through stratified random sampling based on major. Results showed that both AI self-efficacy and AI anxiety were significantly above the neutral midpoint (M = 4.48, SD = 0.76; M = 4.35, SD = 0.85). AI self-efficacy was strongly and negatively associated with AI anxiety (r = - 0.59, p <0.01) and remained a significant negative predictor after controlling for gender, age, and major (β = - 0.112, p = .026). Female teachers reported higher anxiety and lower self-efficacy than male teachers, whereas computer science teachers reported the highest self-efficacy and the lowest anxiety. These findings suggest that university teachers may feel simultaneously capable of using AI and apprehensive about its broader implications. The study provides evidence from Chinese higher education and highlights the value of differentiated institutional support that addresses both teachers' confidence in using AI and their professional concerns.
Pneumoconiosis, a common occupational lung illness, arises from inhaling dust, with silicosis specifically caused by fine crystalline silica dust, leading to lung scarring and inflammation. The diagnosis of silicosis relies on routine monitoring, which includes physical examinations, medical history reviews, and imaging. Chest radiography is a common form of medical screening due to its affordability, efficiency, and suitability for routine use. Recent success in deep learning (DL) for medical image classification has demonstrated that DL algorithms can identify silicosis with high precision by classifying CT images. DL models, specifically convolutional neural networks, have become an effective approach due to their ability to analyse medical images. Therefore, this study presents a novel Multimodal Deep Learning Framework for Early Identification of Silicosis Diagnosis (MDLF-EISD) using radiological images, focused on enabling timely clinical intervention and improving patient outcomes. The proposed framework applies feature fusion (EfficientNet-B3, a capsule network, and ConvNext V2) for integrating complementary radiographic representation, enhancing the ability of the model in capturing disease-specific patterns over different levels of silicosis. Moreover, a convolutional bidirectional attention model is utilised to classify silicosis into corresponding categories effectively. An extensive simulation studies were carried out to evaluate the enhanced performance of the MDLF-EISD method under Silicodata. The comparative analysis of the MDLF-EISD method illustrated a superior accuracy value of 98.73% over other models. These results demonstrate that the feature fusion improves discriminative capabilities for silicosis-related radiographic findings. The proposed system has the ability to support as a computer-aided screening tool for early silicosis diagnosis, particularly in resource-limited clinical and occupational health settings where access to expert radiologists is limited.
Accurate and timely assessment of consciousness is critical for triage, escalation of care, and patient safety in emergency and hospital settings. However, documentation using the AVPU scale (Alert, Verbal, Pain, Unresponsive) remains inconsistent owing to high workload, subjectivity, and fragmented workflows. This study developed and evaluated Consc.ia, a video-based clinical decision-support platform that automates AVPU inference while preserving clinician oversight and enabling seamless, interoperable documentation through HL7 FHIR. A simulated AVPU dataset comprising 136 videos from 58 healthcare professionals (physicians, nurses, paramedics, and first responders) was created under controlled conditions with ethics approval from the ISCTE - Instituto Universitário de Lisboa Ethics Commission (reference CE-ISTA/2025.08, July 2025). The system architecture combines edge-computing computer vision for real-time extraction of facial landmarks, eye state, arm movement, and verbal responses; a clinician-in-the-loop validation layer; and FHIR-mapped Observation resources for direct EHR integration. Three deployment scenarios (Emergency Medical Services, Emergency Departments, and Intermediate Care wards) were designed and compared. Technology adoption was modelled using Rogers' Innovation Adoption Curve and the Bass Diffusion Model (p = 0.01, q = 0.35, M = 111 Portuguese hospitals). The architecture achieves low-latency inference with privacy-by-design (local processing, no raw video storage). Stakeholder validation confirmed strong workflow fit and highlighted persistent documentation gaps during EMS-to-hospital transitions. Scenario analysis revealed distinct hardware and integration requirements (ambulance edge device versus ward multi-camera server). Bass modelling projects gradual adoption, reaching approximately 50% of Intermediate Care wards by 2037 in the realistic scenario, with the "chasm" phase occurring between 2030 and 2032. Sensitivity analysis identified early clinical evidence and FHIR integration support as the strongest accelerators of diffusion. As this constitutes a proof-of-concept study, no quantitative AVPU classification metrics (e.g., accuracy, sensitivity, specificity, or confusion matrix) are reported at this stage; empirical model evaluation against expert-annotated clinical recordings is identified as the primary prerequisite for future validation and clinical translation. As a proof-of-concept that has not yet undergone clinical validation, Consc.ia offers a feasible, interoperable solution for standardising AVPU documentation and strengthening early warning systems. By combining video analytics, edge computing, clinician validation, and FHIR integration, the platform addresses a longstanding gap in emergency-care digitalisation and provides a clear roadmap for real-world adoption.
Traditional micro- and macrodissection techniques enable the extraction of localized regions in thin tissue sections for molecular analysis. Despite the growing use of three-dimensional (3D) microscopy, analogous methods for volumetric microdissection are lacking. Here we have developed a 3D microdissection method based on computer numerical controlled milling integrated with open-top light-sheet microscopy. We demonstrate the ability to study tumor evolution along convoluted 3D branching architectures, which is inaccessible to two-dimensional methods.
Accurate kidney ultrasound segmentation is fundamental for clinical measurement and computer-aided diagnosis. However, domain shifts across devices and centers-manifested as differences in grayscale intensity, contrast, and speckle texture statistics-can substantially degrade model generalization, while acquiring new pixel-level annotations is costly. To address this, we propose a statistical spectral-similarity-guided ultrasound-to-ultrasound translation method to improve kidney segmentation performance without target-domain annotations. Motivated by frequency-domain analysis of renal ultrasound data, we observe that mid-to-low frequency components, which encode global organ structure, exhibit high consistency across domains, whereas mid-to-high frequency components, dominated by device-dependent speckle and texture statistics, vary substantially. Based on dataset-level frequency statistics, our method automatically identifies spectrally similar frequency bands shared by the source and target domains and derives structural guidance from them. This guidance is injected as a soft condition throughout a diffusion-based image generation process, enabling translation to target-device appearance while preserving anatomical structure. The translated images, paired with source-domain labels, are then used to train a segmentation network without requiring any target-domain annotations. Experiments on two public renal ultrasound datasets (OKUS and UNK) and an in-house multi-center dataset demonstrate superior structural preservation in image translation and consistently improved downstream segmentation performance, with particularly large reductions in boundary error. In the challenging OKUS to UNK adaptation scenario, our method boosts the mean Dice score by up to 20.52% (from 56.05% to 76.57%) and drastically reduces the 95% Hausdorff Distance (HD95) boundary error by 71.96 mm compared to the direct transfer baseline. Furthermore, consistent performance gains are achieved across the in-house multi-center dataset. These results indicate that the proposed spectral-similarity-based guidance effectively handles ultrasound domain shifts, substantially improving robustness and generalization for kidney segmentation under zero-shot and cross-center settings.
Patients with multiple myeloma undergoing dorsal spinal instrumentation for malignant spinal lesions remain at high risk for postoperative complications and limited survival. Acute kidney injury (AKI) is common in myeloma and after major surgery, but its prognostic relevance in surgically treated myeloma patients and its association with subsequent CT-based body composition and bone density trajectories, remains insufficiently defined. We performed a retrospective cohort study of consecutive multiple myeloma patients undergoing dorsal spinal instrumentation between 2011 and 2024 at a tertiary referral center. Postoperative AKI was defined and staged according to KDIGO criteria based on serum creatinine changes within 7 postoperative days. Clinical outcomes included overall survival (OS), surgical site infection (SSI), and length of hospital stay (LOHS). CT-based morphometry was assessed on non-contrast whole-body CT at L3 level on a preoperative baseline scan (tCT1) and a postoperative follow-up scan. The follow-up CT scan (tCT2) was obtained approximately 9 months postoperatively as part of routine oncologic follow-up or clinical indication rather than a fixed imaging schedule, reflecting real-world clinical practice. The mean interval between tCT1 and tCT2 was 9.1 ± 1.2 months. Analysis included skeletal muscle index (SMI), skeletal muscle density (SMD), visceral adipose tissue (VAT), and vertebral trabecular bone status assessed by Hounsfield Units (HU). Multivariable Cox regression, logistic regression, and log-linear regression were used to evaluate the independent association of AKI with OS, SSI, and LOHS, adjusting for clinically relevant covariates. 59 patients were included (median age 69.0 years; 40.7% female); postoperative AKI occurred in 16 patients (27.1%). AKI was associated with significantly worse OS (median 224 vs. 396 days without AKI; log-rank p = 0.01), with progressively shorter OS across KDIGO stages. In multivariable Cox regression, AKI remained independently associated with worse OS (adjusted hazard ratio 2.35, 95% CI 1.22-4.54; p = 0.011). AKI was also associated with higher SSI rates (63% vs. 12%; p < 0.01) and longer LOHS (median 29 (IQR 8) vs. 19 (IQR 9) days; p < 0.001). After adjustment for age, sex, preoperative ECOG, and Charlson Comorbidity Index, AKI remained independently associated with higher SSI rates (adjusted odds ratio 3.20, 95% CI 1.11-9.26; p = 0.031) and prolonged hospitalization (LOS ratio 1.34, 95% CI 1.06-1.69; p = 0.014). Longitudinal CT analyses demonstrated significantly greater postoperative declines in the AKI group versus no AKI for SMI (median - 44.6% vs. -18.5%; p < 0.001), SMD (- 20.5% vs. -9.2%; p = 0.02), VAT (- 29.1% vs. -24.1%; p < 0.001), and HU (- 46.2% vs. -37.7%; p < 0.001). In multiple myeloma patients undergoing dorsal spinal instrumentation, postoperative AKI is independently associated with reduced survival, increased postoperative morbidity, and accelerated loss of muscle quantity, muscle quality, visceral fat, and vertebral bone density. These findings highlight AKI as a clinically meaningful systemic event with downstream catabolic consequences and support intensified perioperative nephroprotective and multidisciplinary supportive strategies in this high-risk population.
Interstellar objects provide the only directly observable samples of icy planetesimals formed around other stars, and can therefore provide insight into the diversity of physical and chemical conditions occurring during exoplanet formation1-3. Here we report isotopic measurements of the interstellar comet 3I/ATLAS, which reveal an elemental composition unlike any Solar System body. The water in 3I/ATLAS is enriched in deuterium, at a level of D/H = (0.98 ± 0.06)%, which is more than an order of magnitude higher than in known comets, while its range of 12C/13C ratios (141-191 for CO2 and 123-172 for CO) exceeds typical values found in the Solar System, as well as nearby interstellar clouds and protoplanetary disks. Such extreme isotopic signatures indicate formation at temperatures  ≲ 30 K in a relatively metal-poor environment. When interpreted with respect to models for Galactic chemical evolution, the carbon isotopic composition implies that 3I/ATLAS may have accreted as long ago as 12 billion years, following a period of intense, early star formation. 3I/ATLAS thus represents a preserved fragment of an ancient planetary system.
Coronary artery disease(CAD) is a serious health issue worldwide. Early identification of CAD is used to prevent several complications, such as myocardial infarction and unexpected death. In existing studies, InceptionV3 is computationally intensive and struggles with long-range dependencies, whereas MobileNetV2 faces challenges in extracting intricate features from medical-image data. Similarly, U-NetR, despite its transformer-based encoding, requires large datasets for optimal performance and is computationally expensive because of its self-attention mechanism. To overcome these limitations, this study focuses on merging InceptionV3, U-NetR, and MobileNetV2 to enhance CAD classification performance. This approach involves utilizing pre-trained models and fine-tuning them using an angiographic dataset. The hybrid IUM model incorporates dynamic weighting to maximize prediction accuracy. Furthermore, this study employed VS Grad-CAM visualization to elucidate the classifier decisions using precise heatmaps, thereby improving interpretability. This method achieved exceptional diagnostic metrics: 0.97 accuracy, 0.99 F1-score, 0.98 specificity, and 0.97 sensitivity. This novel approach enhances diagnostic precision, minimizes manual errors, and facilitates real-time applications, making it a scalable and efficient solution for clinical application. Its prompt and accurate identification of CAD has the potential to enhance patient outcomes and optimize healthcare.
Human and animal electrophysiology often looks noisy and unstructured. While some clearer features exist in the form of bursty oscillations and event-related potentials, recent methodological developments have helped to uncover surprising structure in what otherwise looks like noise. This signal, referred to as aperiodic activity, has received substantial recent attention, examining its physiological generators and relationship to behaviour, cognition, development, ageing and disease. Here we examine the putative physiological basis of aperiodic activity, its relationship to other measures of neural activity and evidence for its functional and clinical relevance. Computational modelling and empirical evidence show that aperiodic activity has many neural origins, primarily postsynaptic transmembrane currents across populations of neurons. While several other signal statistics capture features similar to aperiodic measures, new methods have allowed researchers to more directly link aperiodic activity to underlying physiological processes. We discuss the implications of these findings for our understanding of cognition and disease, and highlight open questions for future research.
Biodiversity loss in the present era requires new tools for studying nonmodel organisms. Elephants are both an endangered species and excellent models for studying complex phenotypes including size, social behavior and longevity. Here we report the first derivation of elephant (Elephas maximus) induced pluripotent stem (emiPS) cells. We achieved emiPS cells using two approaches: (1) a two-step process of chemical media induction and colony selection followed by over-expression of elephant transcription factors; and (2) a one-step process with transcription factors and HRAS mutant, HRASG12V. For both protocols, we inhibited TP53 retrogenes, which are hypothesized to confer unique cancer resistance in elephants. To confirm their reprogrammed state, we generated a functional omics catalog of emiPS cells. While these emiPS cells remain transgene-dependent, we inactivated the transgenes and differentiated emiPS cells into all three germ layers via tri-lineage differentiation, embryoid body generation and direct differentiation into putative cell types from all three layers. These methods will open new frontiers for cellular models of nonmodel organisms, including for genetic rescue and conservation.
Nuclear factor-kappa B (NF-κB) is a key transcription factor implicated in inflammation, immune regulation, and cancer progression, making it an important target for antioxidant and anti-inflammatory therapy for acne. The present study evaluated the synergistic NF-κB inhibitory potential of cinnamic acid and p-coumaric acid (p-CA) through molecular docking analysis, followed by formulation development and antioxidant assessment. Molecular docking was performed using AutoDock Vina v1.2.6 to investigate binding affinity and interaction profiles. Individual formulations containing cinnamic acid (1%) and p-CA (1%), as well as an equimolar combined formulation (0.5% each), were developed using hydrogel and oleogel phases to obtain a bigel system. Antioxidant activity was determined using the DPPH radical scavenging assay. Docking studies demonstrated binding energies of - 4.046 kcal/mol and - 4.400 kcal/mol for cinnamic acid and p-CA, respectively, whereas the combined ligand complex exhibited an enhanced binding affinity of - 7.837 kcal/mol. The improved interaction was stabilized through hydrogen bonding and hydrophobic interactions involving key amino acid residues, including ARG54, LEU251, GLU341, and THR342. In the antioxidant assay at 250 μg/mL, p-CA and cinnamic acid exhibited 16.76% and 14.02% inhibition, respectively, IC50 value of the cinnamic acid, p-coumaric acid and combined was found to be 750 µg/mL, 1160 µg/ mL, 810 µg/mL, while the equimolar bigel formulation demonstrated significantly higher radical scavenging activity (19.95%-28.55%), suggesting a synergistic effect. This research indicates that the combination of cinnamic acid and p-CA enhances molecular interactions with the NF-κB p50 subunit and improves antioxidant activity compared with the individual compounds. These integrated in silico and experimental results support the potential application of this combination in the development of multi-targeted natural formulations.
We aimed to report real-world outcomes of transarterial radioembolization (TARE) in patients with intrahepatic cholangiocarcinoma (ICC), focusing on different microsphere types and posttreatment dosimetric approaches in relation to dose-response and survival analyses. This multicenter retrospective single-arm cohort study included adult patients with intrahepatic cholangiocarcinoma treated with Y-90 glass or resin microspheres using lobar or segmental approaches between January 2014 and December 2024 across 13 centers. All posttreatment Y-90 Bremsstrahlung single photon emission tomography (SPECT)/ computed tomography (CT) or Y-90 positron emission tomography (PET) images were reviewed. Post-treatment dose estimations were performed centrally by a nuclear medicine physician with 10 years of experience in dosimetric calculations using the VoxelDosimetry tool of Hermia software (Hermes Medical Solutions, Sweden). Mean perfused liver absorbed dose (PLAD), mean tumor absorbed dose (TAD), and mean whole-liver absorbed dose (WLAD) were calculated. Dosimetric analysis for Y-90 SPECT/CT and PET/CT were performed separately. Tumor response was assessed by comparing imaging obtained within 6 weeks before treatment and 2-4 months after treatment. Response categories were grouped as objective response (complete response + partial response) and nonresponse (stable disease + progressive disease) for treated lesions. Hepatotoxicity was graded according to the Common Terminology Criteria for Adverse Events (CTCAE) using serum Aspartate Aminotransferase (AST), Alanine Aminotransferase (ALT), and bilirubin levels within 3 months after TARE. Additionally, albumin-bilirubin (ALBI) scores were calculated before and after treatment, and changes in ALBI grade were analyzed. Initially, data from 194 (110 female, 84 male; mean age 53.8 ± 16.6 years) patients were included. After excluding patients without survival status data, 180 patients were included in the overall survival (OS) analysis. Seventeen patients were excluded from dose-response evaluation due to suboptimal posttreatment Y-90 imaging for dosimetric calculations. Survival analyses were performed on a per-patient basis, whereas dosimetric and response analyses were performed on a per-treatment-session basis. For dose response analysis, a total of 194 TARE sessions in 163 (101 female, 62 male; mean age 51.4 ± 12.6 years) patients were analyzed. Median (min-max) of administered activities were 2.7 (1.4-6.1) GBq for glass microspheres and 1.4 (0.6-1.8) GBq for resin microspheres (p = 0.004). In separate analyses of SPECT and PET studies, higher TAD and PLAD values were observed in sessions analyzed with PET-based dosimetry (for mean TAD; 249.2 ± 26.6 Gy vs 114.8 ± 15.7 Gy, p < 0.001, for mean PLAD; 102.1 ± 10.4 Gy vs 59.4 ± 11.3 Gy, p = 0.031). A significant positive trend between TAD and response category was observed in the resin microsphere sessions using Y-90 PET-based dosimetry (Area Under Curve 0.693, p = 0.028). In the OS analysis of 180 patients (median follow-up 13.3 months, range 7-81), 128 deaths occurred. Medians of OS did not differ between patients treated with glass versus resin microspheres [17.4 (13.2-21.5, 95%CI) vs 16.6 (7.7-25.5, 95%CI) months, p = 0.80]. Among 107 patients included in time to progression (TTP) analysis, no significant difference in medians of TTP was observed between patients treated with glass and resin microspheres [12.1 (7.8-12.4, 95%CI) vs 14.5 (5.6- 14.4, 95%CI) months, p = 0.11]. One patient developed grade 2 hepatotoxicity and irreversible hepatic failure 10 months after a second TARE session due to tumor progression. In addition, cholangitis occurred in 1 patient and gastritis in 3 patients following TARE. This multicenter analysis suggests that TARE may provide encouraging survival outcomes with acceptable toxicity in patients with ICC, including salvage settings. A potential dose-response association was identified in the resin microsphere subgroup evaluated with Y-90 PET-based dosimetry. Given the retrospective multicenter design and heterogeneity of imaging modalities, these findings should be considered hypothesis-generating and require confirmation in prospective studies using standardized posttreatment dosimetry.
Accurate survival prediction in non-small cell lung cancer (NSCLC) requires integrating clinical, radiological, and histopathological data. Multimodal deep learning (MDL) can improve precision prognosis, but small cohorts and missing modalities limit its clinical applicability, as conventional approaches enforce complete-case filtering or imputation. We present a missing-aware multimodal survival framework that combines computed tomography (CT), whole-slide histopathology images (WSI), and structured clinical variables for overall survival modeling in unresectable stage II-III NSCLC. The framework uses foundation models (FMs) for modality-specific feature extraction and a missing-aware encoding strategy that enables intermediate multimodal fusion under naturally incomplete modality profiles. By design, the architecture processes all available data without dropping patients during training or inference. Intermediate fusion outperforms unimodal baselines and both early and late fusion strategies, with the trimodal configuration reaching a C-index of 74.42. Modality-importance analyses show that the fusion model adapts its reliance on each data stream according to representation informativeness, shaped by the alignment between FM pretraining objectives and the survival task. The learned risk scores produce clinically meaningful stratification of disease progression and metastatic risk, with statistically significant log-rank tests across all modality combinations, supporting the translational relevance of the proposed framework.
In growing 6G-enabled Internet of Vehicles (IoV) environments, in-car networks (IVNs) are more susceptible to sophisticated cyber threats, especially zero-day attacks that avoid signature-based detection. The centralised data dependency, lack of geographical awareness, and poor generalisation in heterogeneous and non-IID conditions are the limitations of current intrusion detection systems. This paper suggests a cooperative intrusion detection system that combines a spatio-temporal graph convolutional long short-term memory (ST-GConvLSTM) model with hierarchical personalised federated learning. The suggested method uses dynamic vehicle-to-vehicle graph structures to capture both intra-vehicle temporal patterns of CAN communications and inter-vehicle spatial dependencies. Scalable training is made possible by a three-tier learning architecture (vehicle, edge, and cloud) that protects data privacy and reduces concept drift. Furthermore, under limitations of latency, energy consumption, communication overhead, and privacy budget, a multi-objective reinforcement learning technique is used to dynamically optimise client involvement. In comparison to state-of-the-art federated baselines, experimental evaluations on real-world CAN datasets under realistic 6G-IoV settings show that the proposed framework achieves high detection performance for both known and zero-day attacks while significantly improving convergence speed and lowering system overhead. These findings demonstrate how well hierarchical federated optimisation and spatiotemporal graph learning can be combined to create safe and scalable vehicular networks.
Achieving spatiotemporal control of light at subwavelength and subcycle scales is an important milestone in the development of new photonic materials for signal processing, pulse shaping and ultrafast imaging. Spatiotemporal light modulation currently relies on electronic interband and intraband transitions that yield pronounced refractive index changes but typically suffer from slow, picosecond response times due to carrier relaxation. Here we show that by leveraging resonant light-matter interactions in a high-quality factor metasurface it is possible to use the optical Kerr effect, a weaker but subfemtosecond optoelectronic polarization effect, to achieve ultrafast, reconfigurable light modulation. By the subwavelength all-optical tuning of the refractive index of the dielectric metasurface unit cells with a spatially structured pump beam, we experimentally demonstrate pulse-limited beam steering with a 74-fs response time at angles up to ±13° in the near-infrared, where the deflection angles are programmable by the pump pattern. The steering originates from the Kerr effect, with a background contribution arising from a slower two-photon-excited free carrier absorption. Additionally, we observe pump self-modulation and self-diffraction, linear frequency conversion, and demonstrate arbitrary subpicosecond spatial light modulation in two dimensions.
PIWI-interacting RNAs (piRNAs) are an important class of non-coding RNA molecules in epigenetic regulation. It plays a crucial role in maintaining genomic stability and inhibiting transposable elements, and have been proven to participate in various diseases by regulating gene expression and influencing signaling pathways. Traditional biological experimental methods have limitations such as low throughput, long cycles, and high costs, making them difficult to meet the requirements of large-scale systematic screening. In this study, we develop a predictive framework named PiDA-DVLSA. We integrate autoencoder, dual graph transformer, and multi-head self-attention mechanisms, and construct an end-to-end multimodal deep learning system. We use autoencoder to perform nonlinear dimensionality reduction and denoising on piRNA sequence features and disease phenotype semantic features, and extract potential representations with strong discriminative ability. Then, we use graph transformers to model the high-order topological relationships between nodes in isomorphic similar graphs, and input heterogeneous graph transformers to learn complex cross-entity interaction patterns in heterogeneous networks. Finally, we achieve adaptive fusion of multi-source information through multi-head self-attention mechanisms. PiDA-DVLSA performs excellently on the benchmark dataset, with AUC and AUPR reach 0.9437 and 0.9195, respectively, significantly outperform eight mainstream algorithms. In independent case validations for breast cancer, clioblastoma, and Alzheimer disease, our model successfully predicts multiple biologically significant potential associations, further confirming its practicality and effectiveness in real scientific research scenarios and providing a solid computational basis for future precision diagnostic and therapeutic applications. PiDA-DVLSA is freely available at https://github.com/zhaoqi106/PiDA-DVLSA .
暂无摘要（点击查看详情）
Intracerebral hemorrhage (ICH) remains associated with high mortality and treatment variability. Current workflows rely on fragmented imaging interpretation and operator-dependent surgical planning. The objective was to develop and validate an agentic artificial intelligence (AI) framework integrating automated imaging analysis, guideline-based reasoning, and trajectory optimization for ICH treatment. Fifty consecutive computed tomography (CT) and computed tomography angiography (CTA) datasets from patients with spontaneous ICH were retrospectively analyzed. The system performed multi-class anatomical segmentation of skin, skull, brain, ventricles, and hematoma, followed by volumetric quantification and JavaScript Object Notation (JSON) based structured encoding of imaging biomarkers. A knowledge-based module incorporating international ICH guidelines generated risk stratification and treatment recommendations. When evacuation was indicated, an automated trajectory modeling module proposed a patient-specific minimally invasive surgical corridor. Overall agreement between AI-generated and expert treatment recommendations was 82% (41/50 cases), with substantial agreement beyond chance (Cohen's κ = 0.71). Discrepancies occurred primarily in borderline surgical indication scenarios. In evacuation candidates, the automated planner generated feasible trajectories in all 50 cases. Median angular deviation between AI-generated and expert-defined trajectories was 7.6°, interquartile range (IQR) 5.1-9.8°. AI-generated trajectories demonstrated equal or greater safety margins relative to expert planning in the majority of cases. End-to-end processing has a potential to substantially reduce simulated decision-support time compared with manual workflow. The proposed agentic AI framework enables structured, explainable, and workflow-integrated decision support for ICH management. This system may reduce operator variability and enhance precision in minimally invasive evacuation planning.
As working memory (WM) is limited in capacity, neural resources must be directed towards task-relevant, and away from task-irrelevant information. Alpha oscillations (8-12 Hz) have been implicated in distractor inhibition during WM retention in younger adults, but it is unclear if alpha oscillations also support distractor inhibition in older adults. We recorded electroencephalography while 24 younger (aged 18-35) and 24 older (aged 60-86) adults completed a modified delay match-to-sample task where distractors of varying strength appeared during the retention period. We found: (1) strong distractors impaired WM performance in both age groups, but only older adults were impaired by weak distractors, relative to the no distractor condition; (2) younger adults demonstrated robust increases in alpha power from baseline during retention, while older adults showed little evidence for a change from fixation; (3) both age groups showed lower alpha power when anticipating distractors compared to no-distractor trials, but younger adults more strongly modulated this response based on distractor strength; and (4) higher pre-distractor alpha power was associated with better task performance in younger adults. Our results suggest age-related differences in both WM retention strategies and anticipatory distractor processing, each of which may contribute to older adults' greater susceptibility to distraction during visual WM.