Agglutinate particles, an important component resulting from micrometeoroids impacts, account for about 13.4% to 84.7% of the volume of lunar regolith depending on its maturity. They are crucial in the soil's evolution and the migration of volatile substances. Here, we examined a representative agglutinate particle from Chang′e-5 samples and modeled how volatiles move through its porous framework. Our analysis revealed that the agglutinate's surface features a patchy distribution of smooth, open pores, as shown by both surface and 3-dimensional structural assessments. By integrating elemental distribution data, we propose that the formation of these smooth, open pores is primarily due to the flow of gaseous volatiles, byproducts of intricate physiochemical reactions occurring in the lunar surface layer during impacts by micrometeoroids. Numerical models of volatile transport in the porous agglutinate have been developed for different flow regimes. These models demonstrate that under the intense conditions of impacts, the transport of volatiles occurs at a remarkably high velocity. Consequently, it is improbable that water would accumulate within the porous structure of lunar soil agglutinates. Nevertheless, understanding this process is valuable for gaining a deeper understanding of the lunar regolith's development and for potential future endeavors in extracting water from the lunar surface.

Takeover safety draws increasing attention in the intelligent transportation as the new energy vehicles with cutting-edge autopilot capabilities vigorously blossom on the road. Despite recent studies highlighting the importance of drivers' emotions in takeover safety, the lack of emotion-aware takeover datasets hinders further investigation, thereby constraining potential applications in this field. To this end, we introduce ViE-Take, the first Vision-driven (Vision is used since it constitutes the most cost-effective and user-friendly solution for commercial driver monitor systems) dataset for exploring the Emotional landscape in Takeovers of autonomous driving. ViE-Take enables a comprehensive exploration of the impact of emotions on drivers' takeover performance through 3 key attributes: multi-source emotion elicitation, multi-modal driver data collection, and multi-dimensional emotion annotations. To aid the use of ViE-Take, we provide 4 deep models (corresponding to 4 prevalent learning strategies) for predicting 3 different aspects of drivers' takeover performance (readiness, reaction time, and quality). These models offer benefits for various downstream tasks, such as driver emotion recognition and regulation for automobile manufacturers. Initial analysis and experiments conducted on ViE-Take indicate that (a) emotions have diverse impacts on takeover performance, some of which are counterintuitive; (b) highly expressive social media clips, despite their brevity, prove effective in eliciting emotions (a foundation for emotion regulation); and (c) predicting takeover performance solely through deep learning on vision data not only is feasible but also holds great potential.

Cell therapy is a promising strategy for acute liver failure (ALF), while its therapeutic efficacy is often limited by cell loss and poor arrangement. Here, inspired by liver microunits, we propose a novel spatially ordered multicellular lobules for the ALF treatment by using a microfluidic continuous spinning technology. The microfluidics with multiple microchannels was constructed by assembling parallel capillaries. Sodium alginate (Alg) solution encapsulating human umbilical vein endothelial cells (HUVECs), hepatocytes, and mesenchymal stem cells (MSCs) are introduced into the middle channel and the 6 parallel outer channels of the microfluidics, respectively. Simultaneously, Ca²⁺-loaded solutions are pumped through the innermost and outermost channels, forming a hollow microfiber with hepatocytes and MSCs alternately surrounding the HUVECs. These microfibers could highly resemble the cord-like structure of liver lobules, bringing about outstanding liver-like functions. We have demonstrated that in ALF rats, our biomimetic lobules can effectively suppress excessive inflammatory responses, decrease cell necrosis, and promote regenerative pathways, leading to satisfied therapeutic efficacy. These findings underscore the potential of spatially ordered multicellular microfibers in treating related diseases and improving traditional clinical methods.

The metaverse enables immersive virtual healthcare environments, presenting opportunities for enhanced care delivery. A key challenge lies in effectively combining multimodal healthcare data and generative artificial intelligence abilities within metaverse-based healthcare applications, which is a problem that needs to be addressed. This paper proposes a novel multimodal learning framework for metaverse healthcare, MMLMH, based on collaborative intra- and intersample representation and adaptive fusion. Our framework introduces a collaborative representation learning approach that captures shared and modality-specific features across text, audio, and visual health data. By combining modality-specific and shared encoders with carefully formulated intrasample and intersample collaboration mechanisms, MMLMH achieves superior feature representation for complex health assessments. The framework's adaptive fusion approach, utilizing attention mechanisms and gated neural networks, demonstrates robust performance across varying noise levels and data quality conditions. Experiments on metaverse healthcare datasets demonstrate MMLMH's superior performance over baseline methods across multiple evaluation metrics. Longitudinal studies and visualization further illustrate MMLMH's adaptability to evolving virtual environments and balanced performance across diagnostic accuracy, patient–system interaction efficacy, and data integration complexity. The proposed framework has a unique advantage in that a similar level of performance is maintained across various patient populations and virtual avatars, which could lead to greater personalization of healthcare experiences in the metaverse. MMLMH's successful functioning in such complicated circumstances suggests that it can combine and process information streams from several sources. They can be successfully utilized in next-generation healthcare delivery through virtual reality.

Gain-of-function mutations of Notch2 cause the rare autosomal dominant disorder known as Hajdu–Cheney syndrome (HCS). Most patients with HCS develop congenital heart disease; however, the precise mechanisms remain elusive. Here, a murine model expressing the human Notch2 intracellular domain (hN2ICD) in cardiomyocytes (hN2ICD-Tg^CM) was generated and the mice spontaneously developed ventricular diastolic dysfunction with preserved ejection fraction and cardiac hypertrophy. Ectopic hN2ICD expression promoted cardiomyocyte hypertrophy by suppressing adenylosuccinate lyase (ADSL)-mediated adenosine 5′-monophosphate (AMP) generation, which further enhanced the activation of the mammalian target of rapamycin complex 1 pathway by reducing AMP-activated kinase activity. Hairy and enhancer of split 1 silencing abrogated hN2ICD-induced cardiomyocyte hypertrophy by increasing Adsl transcription. Importantly, pharmacological activation of AMP-activated kinase ameliorated cardiac hypertrophy and dysfunction in hN2ICD-Tg^CM mice. The frameshift mutation in Notch2 exon 34 (c.6426dupT), which causes early-onset HCS, induces AC16 human cardiomyocyte hypertrophy through suppressing ADSL-mediated AMP generation. Thus, targeting Notch2-mediated purine nucleotide metabolism may be an attractive therapeutic approach to heart failure treatment.

Pre-trained large language models (LLMs) exhibit powerful capabilities for generating natural text. Evolutionary algorithms (EAs) can discover diverse solutions to complex real-world problems. Motivated by the common collective and directionality of text generation and evolution, this paper first illustrates the conceptual parallels between LLMs and EAs at a micro level, which includes multiple one-to-one key characteristics: token representation and individual representation, position encoding and fitness shaping, position embedding and selection, Transformers block and reproduction, and model training and parameter adaptation. These parallels highlight potential opportunities for technical advancements in both LLMs and EAs. Subsequently, we analyze existing interdisciplinary research from a macro perspective to uncover critical challenges, with a particular focus on evolutionary fine-tuning and LLM-enhanced EAs. These analyses not only provide insights into the evolutionary mechanisms behind LLMs but also offer potential directions for enhancing the capabilities of artificial agents.

Cerebral small vessel disease (SVD) involves ischemic white matter damage and choroid plexus (CP) dysfunction for cerebrospinal fluid (CSF) production. Given the vascular and CSF links between the eye and brain, this study explored whether retinal vascular morphology can indicate cerebrovascular injury and CP dysfunction in SVD. We assessed SVD burden using imaging phenotypes like white matter hyperintensities (WMH), perivascular spaces, lacunes, and microbleeds. Cerebrovascular injury was quantified by WMH volume and peak width of skeletonized mean diffusivity (PSMD), while CP volume measured its dysfunction. Retinal vascular markers were derived from fundus images, with associations analyzed using generalized linear models and Pearson correlations. Path analysis quantified contributions of cerebrovascular injury and CP volume to retinal changes. Support vector machine models were developed to predict SVD severity using retinal and demographic data. Among 815 participants, 578 underwent ocular imaging. Increased SVD burden markedly correlated with both cerebral and retinal biomarkers, with retinal alterations equally influenced by cerebrovascular damage and CP enlargement. Machine learning models showed robust predictive power for severe SVD burden (AUC was 0.82), PSMD (0.81), WMH volume (0.77), and CP volume (0.80). These findings suggest that retinal imaging could serve as a cost-effective, noninvasive tool for SVD screening based on vascular and CSF connections.

Robo-pigeons, a novel class of hybrid robotic systems developed using brain–computer interface technology, hold marked promise for search and rescue missions due to their superior load-bearing capacity and sustained flight performance. However, current research remains largely confined to laboratory environments, and precise control of their flight behavior, especially flight altitude regulation, in a large-scale spatial range outdoors continues to pose a challenge. Herein, we focus on overcoming this limitation by using electrical stimulation of the locus coeruleus (LoC) nucleus to regulate outdoor flight altitude. We investigated the effects of varying stimulation parameters, including stimulation frequency (SF), interstimulus interval (ISI), and stimulation cycles (SC), on the flight altitude of robo-pigeons. The findings indicate that SF functions as a pivotal switch controlling the ascending and descending flight modes of the robo-pigeons. Specifically, 60 Hz stimulation effectively induced an average ascending flight of 12.241 m with an 87.72% success rate, while 80 Hz resulted in an average descending flight of 15.655 m with a 90.52% success rate. SF below 40 Hz did not affect flight altitude change, whereas over 100 Hz caused unstable flights. The number of SC was directly correlated with the magnitude of altitude change, enabling quantitative control of flight behavior. Importantly, electrical stimulation of the LoC nucleus had no significant effects on flight direction. This study is the first to establish that targeted variation of electrical stimulation parameters within the LoC nucleus can achieve precise altitude control in robo-pigeons, providing new insights for advancing the control of flight animal–robot systems in real-world applications.

Exosomes (Exos) are emerging as noninvasive biomarkers for diagnosis and progression monitoring of gastric cancer (GC). However, the heterogeneity discrimination and ultrasensitive quantification of Exos presents a considerable analytical challenge, thereby impeding severely their clinical application. Herein, we propose an integrated terahertz metasensing platform for the discrimination of Exos in distinct subtypes of GC in a single step—through the simultaneous evaluation of the category and richness level of Exos membrane proteins. Characterized by dual-sided independent sensing capabilities with enhanced sensitivity (169 and 325 GHz per refractive index unit, respectively), the metasensor functionalized with antibodies simultaneously reflects the content of 2 membrane proteins in the terahertz spectral response. Our approach concurrently completes accurate differentiation and precise quantification of GC-subtype Exos by integrating dual-sided sensing information in merely a single assay. The dual-sided sensing design enhances the reliability of detection results. Moreover, combined with the signal amplification of gold nanoparticles, the platform experimentally demonstrates a superior dynamic response to Exos concentrations spanning from 1 × 10⁴ to 1 × 10⁸ particles/ml, with the limit of detection being 1 × 10⁴ particles/ml. This work provides new insights into multisensing metasurface design and paves the way for precise and personalized cancer treatment through the specific sensing of Exos.