Stochastic computing (SC) has a substantial amount of study on application-specific integrated circuit (ASIC) design for artificial intelligence (AI) edge computing, especially the convolutional neural network (CNN) algorithm. However, SC has little to no optimization on field-programmable gate array (FPGA). Scaling up the ASIC logic without FPGA-oriented designs is inefficient, while aggregating thousands of bitstreams is still challenging in the conventional SC. This research has reinvented several FPGA-efficient 8-bit SC CNN computing architectures, i.e., SC multiplexer multiply-accumulate, multiply-accumulate function generator, and binary rectified linear unit, and successfully scaled and implemented a fully parallel CNN model on Kintex7 FPGA. The proposed SC hardware only compromises 0.14% accuracy compared to binary computing on the handwriting Modified National Institute of Standards and Technology classification task and achieved at least 99.72% energy saving per image feedforward and 31× more data throughput than modern hardware. Unique to SC, early decision termination pushed the performance baseline exponentially with minimum accuracy loss, making SC CNN extremely lucrative for AI edge computing but limited to classification tasks. The SC's inherent noise heavily penalizes CNN regression performance, rendering SC unsuitable for regression tasks.

Early detection and treatment of congenital heart disease (CHD) can significantly improve the prognosis of children. However, inexperienced sonographers often face difficulties in recognizing CHD through transthoracic echocardiogram (TTE) images. In this study, 2-dimensional (2D) and Doppler TTEs of children collected from 2 clinical groups from Beijing Children's Hospital between 2018 and 2022 were analyzed, including views of apical 4 chamber, subxiphoid long-axis view of 2 atria, parasternal long-axis view of the left ventricle, parasternal short-axis view of aorta, and suprasternal long-axis view. A deep learning (DL) framework was developed to identify cardiac views, integrate information from various views and modalities, visualize the high-risk region, and predict the probability of the subject being normal or having an atrial septal defect (ASD) or a ventricular septaldefect (VSD). A total of 1,932 children (1,255 healthy controls, 292 ASDs, and 385 VSDs) were collected from 2 clinical groups. For view classification, the DL model reached a mean [SD] accuracy of 0.989 [0.001]. For CHD screening, the model using both 2D and Doppler TTEs with 5 views achieved a mean [SD] area under the receiver operating characteristic curve (AUC) of 0.996 [0.000] and an accuracy of 0.994 [0.002] for within-center evaluation while reaching a mean [SD] AUC of 0.990 [0.003] and an accuracy of 0.993 [0.001] for cross-center test set. For the classification of healthy, ASD, and VSD, the model reached the mean [SD] accuracy of 0.991 [0.002] and 0.986 [0.001] for within- and cross-center evaluation, respectively. The DL models aggregating TTEs with more modalities and scanning views attained superior performance to approximate that of experienced sonographers. The incorporation of multiple views and modalities of TTEs in the model enables accurate identification of children with CHD in a noninvasive manner, suggesting the potential to enhance CHD detection performance and simplify the screening process.

Pixel-level structure segmentations have attracted considerable attention, playing a crucial role in autonomous driving within the metaverse and enhancing comprehension in light field-based machine vision. However, current light field modeling methods fail to integrate appearance and geometric structural information into a coherent semantic space, thereby limiting the capability of light field transmission for visual knowledge. In this paper, we propose a general light field modeling method for pixel-level structure segmentation, comprising a generative light field prompting encoder (LF-GPE) and a prompt-based masked light field pretraining (LF-PMP) network. Our LF-GPE, serving as a light field backbone, can extract both appearance and geometric structural cues simultaneously. It aligns these features into a unified visual space, facilitating semantic interaction. Meanwhile, our LF-PMP, during the pretraining phase, integrates a mixed light field and a multi-view light field reconstruction. It prioritizes considering the geometric structural properties of the light field, enabling the light field backbone to accumulate a wealth of prior knowledge. We evaluate our pretrained LF-GPE on two downstream tasks: light field salient object detection and semantic segmentation. Experimental results demonstrate that LF-GPE can effectively learn high-quality light field features and achieve highly competitive performance in pixel-level segmentation tasks.

Integrated 2-dimensional (2D) photonic devices such as monolayer waveguide has generated exceptional interest because of their ultimate thinness. In particular, they potentially permit stereo photonic architecture through bond-free van der Waals integration. However, little is known about the coupling and controlling of the single-atom guided wave to its photonic environment, which governs the design and application of integrated system. Here, we report the optical coupling of atomically guided waves to other photonic modes. We directly probe the mode beating between evanescent waves in a monolayer 2D waveguide and a silicon photonic waveguide, which constitutes a vertically integrated interferometer. The mode-coupling measures the dispersion relation of the guided wave inside the atomic waveguide and unveils it strongly modifies matter's electronic states, manifesting by the formation of a propagating polariton. We also demonstrated light modulating and spectral detecting in this compact nonplanar interferometer. These findings provide a generalizable and versatile platform toward monolithic 3-dimensional integrated photonics.

The presence of endotoxemia is strongly linked to the development of endothelial dysfunction and disruption of myocardial microvascular reactivity. These factors play a crucial role in the progression of endotoxemic cardiomyopathy. Sepsis-related multiorgan damage involves the participation of the catalytic subunit of DNA-dependent protein kinase (DNA-PKcs). However, whether DNA-PKcs contributes to endothelial dysfunction and myocardial microvascular dysfunction during endotoxemia remains unclear. Hence, we conducted experiments in mice subjected to lipopolysaccharide (LPS)-induced endotoxemic cardiomyopathy, as well as assays in primary mouse cardiac microvascular endothelial cells. Results showed that endothelial-cell-specific DNA-PKcs ablation markedly attenuated DNA damage, sustained microvessel perfusion, improved endothelial barrier function, inhibited capillary inflammation, restored endothelium-dependent vasodilation, and improved heart function under endotoxemic conditions. Furthermore, we show that upon LPS stress, DNA-PKcs recognizes a TQ motif in cofilin2 and consequently induces its phosphorylation at Thr²⁵. Phosphorylated cofilin2 shows increased affinity for F-actin and promotes F-actin depolymerization, resulting into disruption of the endothelial barrier integrity, microvascular inflammation, and defective eNOS-dependent vasodilation. Accordingly, cofilin2-knockin mice expressing a phospho-defective (T25A) cofilin2 mutant protein showed improved endothelial integrity and myocardial microvascular function upon induction of endotoxemic cardiomyopathy. These findings highlight a novel mechanism whereby DNA-PKcs mediates cofilin2^Thr25 phosphorylation and subsequent F-actin depolymerization to contribute to endotoxemia-related cardiac microvascular dysfunction.

2,2′,7,7′-Tetrakis(N,N-di-p-methoxyphenyl)-amine-9,9′-spirobifluorene (Spiro-OMeTAD) represents the state-of-the-art hole-transporting material (HTM) in n-i-p perovskite solar cells (PSCs). However, its susceptibility to stability issues has been a long-standing concern. In this study, we embark on a comprehensive exploration of the untapped potential within the family of spiro-type HTMs using an innovative anisotropic regulation strategy. Diverging from conventional approaches that can only modify spirobifluorene with single functional group, this approach allows us to independently tailor the two orthogonal components of the spiro-skeleton at the molecular level. The newly designed HTM, SF-MPA-MCz, features enhanced thermal stability, precise energy level alignment, superior film morphology, and optimized interfacial properties when compared to Spiro-OMeTAD, which contribute to a remarkable power conversion efficiency (PCE) of 24.53% for PSCs employing SF-MPA-MCz with substantially improved thermal stability and operational stability. Note that the optimal concentration for SF-MPA-MCz solution is only 30 mg/ml, significantly lower than Spiro-OMeTAD (>70 mg/ml), which could remarkably reduce the cost especially for large-area processing in future commercialization. This work presents a promising avenue for the versatile design of multifunctional HTMs, offering a blueprint for achieving efficient and stable PSCs.

Haptic interactions between human and machines are essential for information acquisition and object manipulation. In virtual reality (VR) system, the haptic sensing device can gather information to construct virtual elements, while the haptic feedback part can transfer feedbacks to human with virtual tactile sensation. Therefore, exploring high-performance haptic sensing and feedback interface imparts closed-loop haptic interaction to VR system. This review summarizes state-of-the-art VR-related haptic sensing and feedback techniques based on the hardware parts. For the haptic sensor, we focus on mechanism scope (piezoresistive, capacitive, piezoelectric, and triboelectric) and introduce force sensor, gesture translation, and touch identification in the functional view. In terms of the haptic feedbacks, methodologies including mechanical, electrical, and elastic actuators are surveyed. In addition, the interactive application of virtual control, immersive entertainment, and medical rehabilitation is also summarized. The challenges of virtual haptic interactions are given including the accuracy, durability, and technical conflicts of the sensing devices, bottlenecks of various feedbacks, as well as the closed-loop interaction system. Besides, the prospects are outlined in artificial intelligence of things, wise information technology of medicine, and multimedia VR areas.

Intelligent metasurfaces have garnered widespread attention owing to their properties of sensing electromagnetic (EM) environments and multifunctional adaptive EM wave manipulation. However, intelligent metasurfaces with broadband high optical transparency have not been studied to date, and most of the previous intelligent metasurfaces lack an integrated design for their actuators and sensors, resulting in lower integration levels. This study proposes a novel intelligent metasurface with adaptive EM wave manipulation ability and high optical transparency from visible to infrared bands. This metasurface consists of a transparent and current-controlled reconfigurable metasurface as an actuator by integrating patterned vanadium dioxide (VO₂) into metal-meshed resonant units, transparent broadband microstrip antenna as a sensor, recognition-and-feedback module, and actuator- and sensor-integrated design on the same substrate. The EM-regulating capability of the designed transparent intelligent metasurface is theoretically analyzed using the coupled mode theory, and a prototype metasurface device is fabricated for experimental verification. Simulation and experimental results demonstrate that the metasurface exhibits over 80% normalized transmittance from 380 to 5,000 nm and adaptive EM wave manipulation (reflective strong shielding function with a shielding efficiency of over 24 dB, high transmittance function with a transmission loss of 1.24 dB, and strong absorption function with an absorption of 97%) according to the EM wave power parameters without manual intervention. This study provides an avenue for transparent intelligent metasurfaces with extensive application prospects in areas such as intelligent optical windows, radar enclosures, and communication.

Cuproptosis-based cancer nanomedicine has received widespread attention recently. However, cuproptosis nanomedicine against pancreatic ductal adenocarcinoma (PDAC) is severely limited by cancer stem cells (CSCs), which reside in the hypoxic stroma and adopt glycolysis metabolism accordingly to resist cuproptosis-induced mitochondria damage. Here, we leverage hyperbaric oxygen (HBO) to regulate CSC metabolism by overcoming tumor hypoxia and to augment CSC elimination efficacy of polydopamine and hydroxyethyl starch stabilized copper-diethyldithiocarbamate nanoparticles (CuET@PH NPs). Mechanistically, while HBO and CuET@PH NPs inhibit glycolysis and oxidative phosphorylation, respectively, the combination of HBO and CuET@PH NPs potently suppresses energy metabolism of CSCs, thereby achieving robust tumor inhibition of PDAC and elongating mice survival importantly. This study reveals novel insights into the effects of cuproptosis nanomedicine on PDAC CSC metabolism and suggests that the combination of HBO with cuproptosis nanomedicine holds significant clinical translation potential for PDAC patients.

Circular RNAs (circRNAs) play a critical regulatory role in degenerative diseases; however, their functions and therapeutic applications in intervertebral disc degeneration (IVDD) have not been explored. Here, we identified that a novel circATXN1 highly accumulates in aging nucleus pulposus cells (NPCs) accountable for IVDD. CircATXN1 accelerates cellular senescence, disrupts extracellular matrix organization, and inhibits mitochondrial respiration. Mechanistically, circATXN1, regulated by heterogeneous nuclear ribonucleoprotein A2B1-mediated splicing circularization, promotes progerin translocation from the cell nucleus to the cytoplasm and inhibits the expression of insulin-like growth factor 1 receptor (IGF-1R). To demonstrate the therapeutic potential of circATXN1, siRNA targeting the backsplice junction of circATNX1 was screened and delivered by tetrahedral framework nucleic acids (tFNAs) due to their unique compositional and tetrahedral structural features. Our siRNA delivery system demonstrates superior abilities to transfect aging cells, clear intracellular ROS, and enhanced biological safety. Using siRNA–tFNAs to silence circATXN1, aging NPCs exhibit reduced mislocalization of progerin in the cytoplasm and up-regulation of IGF-1R, thereby demonstrating a rejuvenated cellular phenotype and improved mitochondrial function. In vivo, administering an aging cell-adapted siRNA nucleic acid framework delivery system to progerin pathologically expressed premature aging mice (zmpste24^−/−) can ameliorate the cellular matrix in the nucleus pulposus tissue, effectively delaying IVDD. This study not only identified circATXN1 functioning as a cell senescence promoter in IVDD for the first time, but also successfully demonstrated its therapeutic potential via a tFNA-based siRNA delivery strategy.

Efficient coordination and planning is essential for large-scale multi-agent systems that collaborate in a shared dynamic environment. Heuristic search methods or learning-based approaches often lack the guarantee on correctness and performance. Moreover, when the collaborative tasks contain both spatial and temporal requirements, e.g., as linear temporal logic (LTL) formulas, formal methods provide a verifiable framework for task planning. However, since the planning complexity grows exponentially with the number of agents and the length of the task formula, existing studies are mostly limited to small artificial cases. To address this issue, a new planning paradigm is proposed in this work for system-wide temporal task formulas that are released online and continually. It avoids two common bottlenecks in the traditional methods, i.e., (a) the direct translation of the complete task formula to the associated Büchi automaton and (b) the synchronized product between the Büchi automaton and the transition models of all agents. Instead, an adaptive planning algorithm is proposed, which computes the product of relaxed partially ordered sets (R-posets) on-the-fly and assigns these subtasks to the agents subject to the ordering constraints. It is shown that the first valid plan can be derived with a polynomial time and memory complexity with respect to the system size and the formula length. Our method can take into account task formulas with a length of more than 400 and a fleet with more than 400 agents, while most existing methods fail at the formula length of 25 within a reasonable duration. The proposed method is validated on large fleets of service robots in both simulation and hardware experiments.

Somatic cell reprogramming generates induced pluripotent stem cells (iPSCs), which serve as a crucial source of seed cells for personalized disease modeling and treatment in regenerative medicine. However, the process of reprogramming often causes substantial lineage manipulations, thereby increasing cellular heterogeneity. As a consequence, the process of harvesting monoclonal iPSCs is labor-intensive and leads to decreased reproducibility. Here, we report the first in-house developed robotic platform that uses a pin-tip-based micro-structure to manipulate radial shear flow for automated monoclonal iPSC colony selection (~1 s) in a non-invasive and label-free manner, which includes tasks for somatic cell reprogramming culturing, medium changes; time-lapse-based high-content imaging; and iPSCs monoclonal colony detection, selection, and expansion. Throughput-wise, this automated robotic system can perform approximately 24 somatic cell reprogramming tasks within 50 days in parallel via a scheduling program. Moreover, thanks to a dual flow-based iPSC selection process, the purity of iPSCs was enhanced, while simultaneously eliminating the need for single-cell subcloning. These iPSCs generated via the dual processing robotic approach demonstrated a purity 3.7 times greater than that of the conventional manual methods. In addition, the automatically produced human iPSCs exhibited typical pluripotent transcriptional profiles, differentiation potential, and karyotypes. In conclusion, this robotic method could offer a promising solution for the automated isolation or purification of lineage-specific cells derived from iPSCs, thereby accelerating the development of personalized medicines.

Catheters navigating through complex vessels, such as sharp turns or multiple U-turns, remain challenging for vascular embolization. Here, we propose a novel multistage vascular embolization strategy for hard-to-reach vessels that releases untethered swimming shape-memory magnetic microrobots (SMMs) from the prior catheter to the vessel bifurcation. SMMs, made of organo-gel with magnetic particles, ensure biocompatibility, radiopacity, thrombosis, and fast thermal and magnetic responses. An SMM is initially a linear shape with a 0.5-mm diameter at 20 °C inserted in a catheter. It transforms into a predetermined helix within 2 s at 38 °C blood temperature after being pushed out of the catheter into the blood. SMMs enable agile swimming in confined and tortuous vessels and can swim upstream using helical propulsion with rotating magnetic fields. Moreover, we validated this multistage vascular embolization in living rabbits, completing 100-cm travel and renal artery embolization in 2 min. After 4 weeks, the SMMs maintained the embolic position, and the kidney volume decreased by 36%.

Fut2-mediated α1,2-fucosylation is important for gut homeostasis, including the intestinal stem cell (ISC). The stemness of ISC declines with age, and aging-associated ISC dysfunction is closely related to many age-related intestinal diseases. We previously found intestinal epithelial dysfunction in some aged Fut2 knockout mice. However, how Fut2-mediated α1,2-fucosylation affects ISC aging is still unknown. On this basis, the herein study aims to investigate the role of Fut2-mediated α1,2-fucosylation in ISC aging. Aging models in ISC-specific Fut2 knockout mice were established. ISCs were isolated for proteomics and N-glycoproteomics analysis. ISC functions and mitochondrial functions were examined in mice and organoids. Ulex europaeus agglutinin I chromatography and site-directed mutagenesis were used to validate the key target fucosylated proteins of Fut2. As a result, Fut2 knockout impaired ISC stemness and promoted aging marker expression in aged mice. Proteomics analysis indicated mitochondrial dysfunction in Fut2 knockout ISC. More injured mitochondria, elevated levels of reactive oxygen species, and decreased levels of adenosine 5′-triphosphate (ATP) in Fut2 knockout ISC were found. Moreover, respiratory chain complex impairment and mitophagy dysfunction in Fut2 knockout ISC were further noted. Finally, Fut2 was demonstrated to regulate mitochondrial functions mainly by regulating the α1,2-fucosylation of N-acyl sphingosine amidohydrolase 2 (Asah2) and Niemann–Pick type C intracellular cholesterol transporter 1 (Npc1). In conclusion, this study demonstrated the substantial role of Fut2 in regulating ISC functions during aging by affecting mitochondrial function. These findings provide novel insights into the molecular mechanisms of ISC aging and therapeutic strategies for age-related intestinal diseases.

Metastasis is the major cause of cancer-related death, and lymph node is the most common site of metastasis in breast cancer. However, the alterations that happen in tumor-draining lymph nodes (TDLNs) to form a premetastatic microenvironment are largely unknown. Here, we first report the dynamic changes in size and immune status of TDLNs before metastasis in breast cancer. With the progression of tumor, the TDLN is first enlarged and immune-activated at early stage that contains specific antitumor immunity against metastasis. The TDLN is then contracted and immunosuppressed at late stage before finally getting metastasized. Mechanistically, B and follicular helper T (Tfh) cells parallelly expand and contract to determine the size of TDLN. The activation status and specific antitumor immunity of CD8⁺ T cells in the TDLN are determined by interleukin-21 (IL-21) produced by Tfh cells, thus showing parallel changes. The turn from activated enlargement to suppressed contraction is due to the spontaneous contraction of germinal centers mediated by follicular regulatory T cells. On the basis of the B-Tfh-IL-21-CD8⁺ T cell axis, we prove that targeting the axis could activate TDLNs to resist metastasis. Together, our findings identify the dynamic alterations and regulatory mechanisms of premetastatic TDLNs of breast cancer and provide new strategies to inhibit lymph node metastasis.