共找到 20 条结果
Whistleblower alleges Finnish startup's vaunted solid-state battery isn't what it claims.
Hyperspectral remote sensing data provides distinct advantages for lithological classification in bedrock-exposed areas. Despite the superior performance of ensemble learning methods (e.g., Rotation Forest, ROF) in big data classification, their application in high-dimensional hyperspectral data is restricted by high training costs. To address this limitation and improve classification accuracy, this study proposes an optimized ROF-LightGBM ensemble algorithm integrated with minimum noise fraction (MNF) for rotation matrix construction. Experimental validation was conducted using ZY1-02D hyperspectral data for lithological mapping in the bedrock-exposed Xitieshan area, involving ROF-LightGBM parameter optimization (L × T, bootstrap) and comparative experiments with multiple machine learning models. The results demonstrate the following: under the same number of decision trees (T = 100), the ROF-LightGBM (PCA, L × T = 4 × 25) with optimized base classifier outperforms random forest (RF), LightGBM, and traditional ROF (L = 100) models in classification accuracy, achieving 74.28% accuracy, 6.54% higher than RF and 1.53% higher than LightGBM. More notably, it boasts exceptional efficiency, with a training time of only 4.86 s (nearly 37 times shorter than traditional ROF), while maintaining minimal accuracy loss (an only 1.19% decrease). Additionally, the ROF-LightGBM (MNF) model, which adopts MNF for rotation matrix construction, further enhances performance. Compared with the PCA-based ROF-LightGBM, it achieves an 82.17% classification accuracy (a 7.89% increase) and its kappa coefficient reaches 0.81, fully verifying the model's superiority in accuracy and efficiency.
The article presents the INSIDE-OUT longitudinal technique for bladder neck closure in children, a simplified approach aimed at improving outcomes in pediatric patients, particularly those with myelomeningocele. Traditionally, these patients undergo bladder neck plasty or sling procedures, which, while potentially preserving bladder function, achieve continence rates of approximately 50 %. In contrast, the new INSIDE-OUT technique boasts a success rate exceeding 90 %. The method begins with a midline incision of the bladder, allowing for internal exposure of the bladder neck. A Foley catheter is inserted and traction applied while plastic tubes are positioned in the ureters to facilitate transection of the bladder neck. The technique involves circumferential dissection of the bladder neck from the inside, creating flaps that are elevated to skin level. Closure occurs in two layers, using specific sutures for mucosal and seromuscular layers. Results show a continence rate exceeding 95 % after over a decade of implementation. We believe that the INSIDE-OUT technique should be more widely considered in complex bladder reconstructions, emphasizing the need to educate patients about the risks associated with improper catheterization. This innovative approach promises better outcomes for children facing complex urinary issues.
Antibiotic tolerance enables populations of microbes to survive normally lethal antibiotic concentrations, increasing the likelihood of reinfection and facilitating the evolution of resistance. Tolerance measurements typically involve quantifying viable cells after antibiotic exposure. Existing methods range from accessible but low-throughput approaches, such as plate counting, to higher-throughput but semi-quantitative techniques, such as the TDtest. Here, we develop a new system for rapid, precise and high-throughput tolerance measurements. We utilize Surface Patterned Omniphobic Tiles (SPOTs) to discretize cell suspensions into nano-to microliter droplets and estimate the viable cell concentrations following antibiotic exposure from the proportion of empty droplets using Poissonian statistics. We apply the platform to monitor Klebsiella pneumoniae tolerance to meropenem over time as a proof of concept. The resulting assay is accessible, compatible with multiple media, and boasts a large dynamic range, sufficient resolution, and rapid handling.
Information bottleneck (IB) theory aims to compact coding while retaining task-relevant information, but the lack of prior knowledge in clustering poses challenges to its application. Besides, since most methods fail to emphasize clustering security, empty clusters caused by excessive preference can lead to significant performance degradation. Motivated by these, a scalable framework is proposed, named Information Bottleneck Inspired Balanced Multiview Clustering (IB2MC), which boasts clustering security, optimization synergy, and tuning simplicity. To begin with, the in-depth discussion on unsupervised IB theory reveals the balance principle via maximizing label entropy, which preserves the uncertainty of cluster distribution to overcome bias or neglect for clusters. On this basis, compressed descriptors are expected to evolve toward discriminability, thus facilitating label inference from them at the same time. Inspired by this, our method proposes dynamic cosine graph learning for label transmission, and achieves seamless label extraction to strengthen dual-level consistency. Moreover, the representation alignment under independent constraint can enhance cluster separability, while the trace ratio for self-balance clustering can further avoid extra hyperparameter tuning. In this way, the time complexity of our method is linear with sample size, while no pre- or post-processing is required for joint optimization. To demonstrate the effectiveness of our method, seventeen state-of-the-art methods are chosen as baselines and our method ranks first in average performance across all twelve real-world data sets.
BACKGROUND: Nurses are integral to delivering humanistic care in palliative care settings. While Artificial Intelligence (AI) boasts significant potential for palliative care practices, its integration raises critical ethical concerns such as AI’s impact on patient autonomy and its risks of depersonalized care that warrant further exploration. METHODS: A qualitative study using reflexive thematic analysis was conducted to explore nurses’ perspectives on the potential impact of AI integration in palliative care, with a focus on ethical considerations and the prospective actions needed for implementation. Semi-structured interviews were conducted with 20 registered nurses experienced in palliative care, recruited from four hospitals across two urban centers in Sichuan Province, China. RESULTS: Nurses perceive AI integration as ethically transformative rather than merely technical, raising concerns regarding autonomy, trust, justice, and cultural alignment. The thematic analysis identified five major ethical considerations regarding AI integration in palliative care from nurses’ perspectives: (1) ethical challenges to patients’ autonomy and dignity; (2) AI-driven reconstruction of trust dynamics; (3) decision making conflicts in palliative care; (4) difficulties in achieving equity in AI-enabled health services; and (5) adaptation to cultural sensitivity. CONCLUSIONS: The inclusion of AI ethics and dynamic moral decision‑making modules in nursing education is suggested to enhance nurses’ moral sensitivity to deliver equitable care, and strengthen their digital literacy and ethical awareness in AI‑mediated environments. Nurses may assume an ethical responsibility to reduce the risks of cultural bias in AI tools, and to advocate the integration of multicultural perspectives into AI design, training and evaluation, thereby advancing the development of more culturally responsive technologies.
Optical imaging is crucial in preclinical evaluation of tumor development, progression, and therapies, due to its low cost, versatility, sensitivity, and capability of real-time monitoring. The main in vivo imaging techniques include fluorescence, bioluminescence, and photoacoustic imaging. Although all three applications are limited by relatively low tissue penetration due to absorption and scattering of light in biological tissues, each application has its own benefits. Fluorescence imaging is technically the simplest method to perform and using near infrared fluorescent (NIRF) probes can enhance depth capability. Moreover, novel activatable probes, lifetime imaging, and multimodal imaging strategies have extended its utility to endoscopic and surgical procedures. Bioluminescence imaging boasts high sensitivity and signal-to-noise ratio, as light emission originates at the target from enzymatic reactions. However, it requires genetic modification to express luciferases, restricting its use in humans. Photoacoustic imaging offers better depth penetration compared to purely optical methods by utilizing ultrasound signals. Leveraging endogenous contrast from chromophores, it assists in vascular imaging and tumor identification without external labels. Due to the use of safe, non-ionizing radiation, it reduces the risk to subjects and allows repeated imaging, and is thus most suited for clinical translation. All three applications continuously benefit from the development of three-dimensional imaging techniques and deep learning algorithms.
Photon-Counting Detector Computed Tomography (PCD-CT) boasts excellent spectral utilization capability. Combined with material decomposition methods, it enables the calculation of the effective atomic number ($Z_{eff}$) and density ($\rho$) of scanned materials. Traditional material decomposition methods derive $Z_{eff}$ and $\rho$ from physical models using either the basis material model or the dual-effect model. However, these methods generally fail to meet high-precision decomposition requirements due to model approximations and suffer from the limitation of severe noise amplification in $Z_{eff}$ images. This study proposes a physics-constrained deep learning network that achieves high-precision joint estimation of $Z_{eff}$ and $\rho$ by dynamically modeling nonlinear X-ray interactions. A Multilayer Perceptron (MLP) is employed to construct dynamic compensation functions dependent on energy ($E$) and $Z_{eff}$ for the photoelectric effect exponent and Compton scattering model in the X-ray interaction model. The decomposition network adopts Swin-Unet as its backbone and utilizes a hybrid loss function, which consists of L1 loss and SSIM loss for $Z_{eff}$/$\rho$, as well as a physics-informed loss derived from the L1 loss between the predicted $Z_{eff}$/$\rho$ and the monoenergetic linear attenuation coefficient images generated by the optimized X-ray model. This design allows the network to simultaneously learn data-driven features and physical principles. Comparative experiments were conducted on a PCD-CT system between the proposed method and four methods (Lan, U-Net, Butterfly-net, and Swin-Unet without physical constraints). The results demonstrate that: for standard materials, the proposed method achieves a Mean Absolute Percentage Error (MAPE) below 5% for both $Z_{eff}$ and $\rho$ decomposition, with superior Noise Power Spectrum (NPS) performance; for biological samples including freshwater crayfish and mouse, the $Z_{eff}$ images generated by the proposed method exhibit higher Multi-Exposure Fusion Structural Similarity Index (MEF-SSIM), reaching 0.9559 for crayfish and 0.8950 for mouse. The method also demonstrates superior detail recovery capability in the restoration of the speckled tissue structure of crayfish and the tissue regions of mouse. By incorporating constraints from monoenergetic images generated based on the dynamic X-ray interaction model, the network effectively learns the nonlinear decomposition process of $Z_{eff}$ within a data-driven framework. The proposed method improves decomposition accuracy while reducing decomposition noise and enhancing the quality of decomposed images.
Genetic mutations in the leucine-rich repeat kinase 2 (LRRK2) protein have been linked to Parkinson's disease (PD), a disabling and progressive neurodegenerative disorder for which treatments are limited. Herein, we describe the invention of a macrocyclic LRRK2 inhibitor lead chemical series. Rigorous application of knowledge-, structure-, and property-based drug design culminated in the discovery of compound 7, which was profiled extensively before it was determined to be clastogenic, which halted its progression. Parallel optimization of kinome selectivity and PXR activation through structure- and property-based drug design resulted in the discovery of the lead macrocycle compound 12. This macrocycle boasts a remarkably low projected human QD dose, is nongenotoxic, and achieved encouraging brain penetration in early preclinical models.
We report a scalable and versatile protocol for fabricating high-performance metal disk ultramicroelectrodes (UMEs) with precisely controlled geometries for multifunctional scanning electrochemical microscopy (SECM). By integrating a thick-wall nanopipette template with tip forging, our approach achieves precise control over probe morphology, yielding dual-disk electrodes (DDE) with RG values of <1.1 and interelectrode gaps of 5-10 μm. This reproducible method boasts a success rate of >95% across large batches. We demonstrate the practical utility of these probes through simultaneous O2 and pH monitoring during water oxidation catalysis and in situ mapping of lithium-ion flux during lithium plating. Our findings establish a reliable platform for the production of sensitive, multifunctional SECM probes, significantly advancing the quantitative investigation of complex interfacial chemical processes in electrocatalysis and energy storage.
QiangHuoShengShi decoction (QHSS), a classical prescription traditional Chinese medicine (TCMs), boasts a lengthy clinical history in treating various rheumatic diseases. No pharmacokinetic (PK) studies on QHSS were reported to date while its in vitro and in vivo chemical fingerprints were comprehensively reported. This study aims to reveal the time-dependent changes in the concentrations of multi-known and -unknown components in rats after oral administration of QHSS, thereby offering insights for further research into its pharmacodynamic mechanisms. The prediction of multiple reaction monitoring (MRM) parameters for known and unknown analytes was first achieved using a standards-independent ultra-high performance liquid chromatography-quadrupole time-of-flight mass spectrometry (UHPLC-Q-TOF-MS/MS) method coupled with linear correlation analysis. Subsequently, specific parameters were then determined using single-factor optimization by the ultra-high performance liquid chromatography-triple quadrupole mass spectrometry (UHPLC-MS/MS). Finally, the concentrations of known and unknown analytes in plasma biosamples were determined using a built scheduled multiple reaction monitoring (sMRM) method. The sMRM parameters for 90 compounds were predicted and optimized. The pharmacokinetic parameters, including mean time to reach peak concentration (Tmax), maximum plasma concentration (Cmax), area under concentration-time curve (AUC), mean resident time (MRT), were characterized for 32 measurable components (17 with standard substances and 15 without standard substances). The pharmacokinetic parameters of 28 components from biosamples of the rheumatoid arthritis (RA) group were further compared with those from the normal group, revealing significant differences (either increased or decreased) in their PK parameters. A pseudo-targeted pharmacokinetic study of QHSS was successfully conducted by a standards-independent UHPLC-Q-TOF-MS/MS with UHPLC-sMRM-MS/MS method, which holds significant promise for subsequent research into the medicinal substances of QHSS. It can also serve as a powerful tool for multi-compounds pharmacokinetic study of more TCMs.
Current methods to assess growth plate predominantly rely on rudimentary metrics such as tissue thickness, which fail to capture the dynamic, spatial and functional heterogeneity of chondrocyte populations during endochondral ossification. To address this gap, we developed a computational histomorphometric pipeline (Growth Plate Professional Analyzer, "GP Pro"), designed to quantify chondrocyte kinetics and maturation gradients at single-cell resolution using routine histological tissue sections. GP Pro integrates three automated modules: (1) batch processing of whole-slide images (WSIs) with growth plate localization, (2) segmentation of the growth plate using Segment Anything Model 2 (SAM2), and (3) single-cell lacunae analysis to extract morphometric and spatial distribution features across differentiation stages. For GP PRO's feasibility study, we conducted high-throughput testing on mouse/hamster/rat knee sections stained with Safranin-O/fastgreen (Safranin-O) and Hematoxylin & Eosin (H&E) to evaluate the computational efficiency and accuracy of the whole system and built-in functions. For GP PRO's application validation, tibial growth plates from normotensive Wistar-Kyoto (WKY) and spontaneously hypertensive rats (SHRs) were tested by the pipeline, correlating outputs with histochemical staining and micro-CT scanning results. GP PRO demonstrates satisfactory computational efficiency, with a processing time as low as 3.6 s per section and achieves an end-to-end automated processing success rate of above 95 %. Additionally, it boasts an average chondrocyte identification accuracy of 90.28 % and a macro-precision of 94.76 % in cell classification. In the application of SHR, GP Pro revealed a 4.03 percentage points increase in proliferative chondrocyte proportion (p < 0.05; 95 % CI: 0.59-7.47) and the presence of aberrantly elongated pre-hypertrophic chondrocytes at 3-month-old, indicative of maturation arrest. These cellular perturbations aligned with micro-CT showing a 36.04 percentage points decrease in primary spongiosa bone volume fraction (p < 0.001; 95 % CI: 30.10-41.97), suggesting compromised osteogenic coupling. Histology confirmed premature hyperproliferation and delayed hypertrophic differentiation in SHR growth plates. GP Pro establishes a scalable platform for mapping chondrocyte behavior in situ, enabling unprecedented resolution of growth plate pathobiology. By linking hypertrophic differentiation delays to trabecular bone deficits in hypertension, this tool advances mechanistic studies of endochondral ossification and offers translational potential for diagnosing growth disorders or monitoring therapeutic interventions in multiple species. This study introduces GP Pro, an AI-driven histomorphometric platform that bridges pre-clinical research and clinical applications by enabling high-throughput, single-cell resolution analysis of growth plate dynamics directly from WSIs.
A facile method for the synthesis of stereochemically well-defined ketonyl cysteine compounds has been established. This was achieved via an N-phenylphenothiazine-photocatalyzed, radical-based desulfurative addition, using the in situ-generated tetrafluoropyridyl cysteine derivative from the thiol as a key precursor, which ensures high retention of configuration. Notably, this method features simple operation: the phenothiazine photosensitizer employed in the reaction is inexpensive and readily available in comparison with costly iridium catalysts. It also boasts the advantages of easily accessible starting materials, mild reaction conditions, excellent functional group tolerance, high stereoretention, and good yields, thus exhibiting broad applicability in the modification of complex peptide molecules.
To analyze the equity and utilization efficiency of health resource allocation in Anhui Province, providing methods for rational resource layout. The entropy-weighted TOPSIS method and rank-sum ratio (RSR) method were adopted to conduct comprehensive evaluation and grading analysis of health resource allocation in Southern, Central, and Northern Anhui from 2019 to 2023. The entropy-weighted TOPSIS analysis indicated that since 2021, the comprehensive evaluation index (C-value) of Central Anhui surpassed that of Southern Anhui, reaching 0.745 in 2023, while Northern Anhui consistently exhibited low C-values. RSR grading results classified Southern Anhui as "excellent," Central Anhui as "moderate," and Northern Anhui as "poor." Significant disparities exist in health resource allocation across Anhui Province. Southern Anhui boasts abundant resources but faces weaker grassroots service capacity; Central Anhui shows rapid growth yet poor alignment between resources and population density; Northern Anhui suffers from inadequate infrastructure and high equipment idle rates. Recommendations include:strengthening mobile medical services and resource allocation to grassroots in Southern Anhui; implementing hierarchical diagnosis and treatment coupled with medical consortium collaboration in Central Anhui; enhancing grassroots talent and equipment management in Northern Anhui through targeted training programs and incentive policies.
As microgrid complexity increases, cyber-physical coordination must be secure and efficient. This research designs an edge-AI blockchain framework for the cyber-physical management of renewable-integrated microgrids. The edge devices call SNNs for low-power real-time learning and fault detection functions; Hyperledger Fabric performs the functions of energy transaction validation and energy access control. The microgrid edge node leverages the SNN to predict faults and perform switching optimization lately on noise and latency. The blockchain guarantees trusted peer-to-peer communication and secure provenance of data. The system boasts Cyber Fault Detection Accuracy (CFDA) of 97.6%, Consensus Delay < 2.3 s and Voltage Deviation < ± 1.1%. The edge-AI + blockchain system reduces communication overhead and enhances energy authentication efficiency by 28% over centralized control. The architecture also allows adaptive restoration mechanisms upon disturbance and integrates hierarchical control across layered microgrid clusters. Edge AI speeds up anomaly detection with minimal computation costs while maintaining grid observability. The unified system is aimed at providing greater cyber resilience and real-time suitability for adverse operational conditions. Simulations performed on MATLAB Simscape and Hyperledger test networks verify that such an arrangement improves system stability, recovers from faults, and extends its controlability to a great extent, thus standing in line to the developing standards and polices for a decentralized microgrid setup.
Radiation detection technology is critical in medical diagnosis, high-energy physics experiments, nuclear environmental monitoring, and radiation safety protection. Its technological iteration stems from innovations in high-performance radiation detection materials. Traditional materials often have narrow dose-response intervals, insufficient high-precision measurement capability, low spatial resolution, and poor stability, failing to meet high-precision detection requirements. Ag-doped phosphate glass (Ag-PG), based on radio-photoluminescence (RPL), effectively addresses these limitations with its comprehensive advantages: high radiation sensitivity, a wide linear dose-response range, submicron spatial resolution for radiation imaging, write-erase-rewrite capability, and visualized dose monitoring potential, and it also boasts significant fundamental research value and engineering application prospects. Specifically, while existing RPL reviews mainly provide a comprehensive analysis from the perspective of RPL and present typical RPL material systems, this paper systematically analyzes the structural characteristics of the Ag-PG matrix and the coordination configuration and site occupation of Ag ions. It clarifies RPL luminescence properties, dose-response mechanisms, and the evolution of luminescence centers, while reviewing advancements in applications such as radiation dose detection and high-resolution X-ray imaging. By summarizing the current research status, technical advantages and existing challenges of Ag-PG, this study provides theoretical references and conceptual insights to promote breakthroughs in its fundamental research and practical applications in high-precision radiation dose detection, advanced medical imaging, micro-nano-scale radiation detection, and nuclear industry non-destructive testing.
Repurposing waste materials that would otherwise be discarded and finding new pathways for them to participate in reactions for recycling are parts of a highly atom-economical and environmentally friendly strategy. Drawing upon this concept, we have developed a method that utilizes Et3SiOBpin, a borasiloxane typically obtained as a byproduct; limited reports have explored its potential applications in organic synthesis, to facilitate the construction of diverse multisubstituted perfluoroalkyl-boryl olefins via tetracoordinate boron species under photoinduced conditions. Despite challenges such as reaction progress, yield, and selectivity, we overcame these challenges by fine-tuning the oxygen, silicon, and alkyl groups in the borasiloxane reagent. This protocol boasts readily available starting materials, high atom economy, a broad substrate scope, and diversified valuable products and, notably, marks the first successful use of borasiloxane as a dual reagent (boron reagent and multifunctional base) in a synthetic methodology. This advancement holds significant potential to enrich and expand synthetic chemistry research.
The L-alanine boasts a plethora of industrial applications; however, its commercial utilization is hindered by high production costs, scarcity of raw materials, compromised fermentative productivity, and low enantiomeric purity. To address these challenges in amino acid biosynthesis, the present study employed a synthetic metabolic engineering strategy to enhance L-alanine production in the industrially robust Escherichia coli BL21 (DE3) strain. A synthetic alaD gene cassette, computationally designed and cloned into the pUC57 vector, was expressed under the control of a high-efficiency T7 promoter. Initial expression in recombinant E. coli BL21 (DE3) (alaD⁺) under unoptimized conditions resulted in modest L-alanine titres of 54.32 mM (4.84 g/L). However, through statistical optimization using Response Surface Methodology (RSM) and cultivation under oxygen-limited batch fermentation, the L-alanine yield increased markedly to 440.47 mM (39.24 g/L) within 24 h. Given its broad industrial applicability, this biosynthetic approach offers a promising and sustainable alternative to conventional chemical synthesis, positioning recombinant E. coli BL21 (DE3) (alaD⁺) as a viable microbial platform for commercial production of L-alanine.
The atomic magnetometer (AM), operating within the spin-exchange relaxation-free (SERF) regime, boasts numerous advantageous qualities, including ultrahigh sensitivity, exceptional spatial resolution, and minimal power consumption. Consequently, it emerges as a promising alternative to superconducting quantum interference devices in biomagnetic measurement applications. This paper details the development of a fully integrated SERF AM system comprising a compact sensor head and corresponding control electronics. Utilizing a 4 mm × 4 mm × 4 mm cubic vapor cell, we have successfully integrated the compact sensor into a 9 cm3 volume employing a single-beam scheme facilitated by a polarization-maintaining fiber. The in-house control electronics encompass essential components, such as the laser driver, coil driver, vapor-cell temperature controller, and transimpedance amplifier. As a result, the fully integrated SERF AM achieves a sensitivity of 25 fT/Hz1/2@5∼100 Hz, accompanied by a bandwidth of 193 Hz, meeting the necessary criteria for magnetocardiography (MCG) and magnetoencephalography (MEG) measurements. Furthermore, the fully integrated SERF AM successfully records typical MCG and alpha rhythm MEG signals, showcasing immense potential for biomagnetic imaging applications.
Herein, we developed a novel protocol for the visible-light-catalyzed synthesis of multifunctional pyrrole-2-ones from N-arylglycines (1) and cyclopropenones (2) through decarboxylation, radical addition, and ring expansion reactions. Furthermore, this cascade reaction requires only 0.08 equivalents of 4CzIPN to facilitate the photocatalytic decarboxylation, free radical addition, and ring expansion reactions, thereby eliminating the need for any transition metal catalyst. Target compounds 3a-3l' were obtained in moderate to excellent yields (57%-96%) by employing 4CzIPN as the photocatalyst, NaHCO3 as the base, and propylene carbonate (PC) as the green solvent under an argon atmosphere with blue light irradiation for 3 h. The reaction does not require any metal catalyst or heating, has a wide substrate scope, is environmentally friendly and sustainable, boasts a high atom economy, and exhibits significant regioselectivity. This strategy facilitated the synthesis of pyrrole-2-ones bearing quaternary centers with potential biological activity, expanding the applications of N-arylglycine as a C2 synthon in the construction of pyrrole-2-one frameworks via room-temperature photochemical reactions.