Search Results (179)

While chimeric antigen receptor (CAR)-T-cell therapies have shown significant effectiveness in hematological malignancies, their efficacy in solid tumors remains limited by the hostile tumor microenvironment (TME) and antigen heterogeneity. Recently, CAR-Macrophage (CAR-M) therapy has emerged as a paradigm-shifting approach, leveraging the innate capability of macrophages to deeply infiltrate tumors and their plasticity to reverse immunosuppression. Unlike T cells, CAR-Ms not only mediate direct phagocytosis but also initiate epitope spreading, effectively bridging innate and adaptive immunity. This review critically examines the trajectory of CAR-M therapy from biological rationale to clinical reality. We dissect the engineering evolution of CAR constructs, arguing for macrophage-specific signaling domains (e.g., FcRγ, Megf10) over traditional T-cell designs. Crucially, we address the major bottlenecks in clinical translation, including the manufacturing challenges of non-expanding primary macrophages and the emerging shift toward induced pluripotent stem cell (iPSC)-derived platforms. Furthermore, we evaluate current clinical trial landscapes and discuss next-generation strategies such as in vivo programming via lipid nanoparticles (LNPs) and synthetic logic-gating to enhance safety. Ultimately, overcoming manufacturing constraints and optimizing delivery systems will be pivotal for CAR-M to evolve from a niche therapy into a standard-of-care modality for solid tumors. Full article

Mechanistic interpretability seeks to reverse-engineer the computational circuits within large language models, but current methods rely on exhaustive or heuristic search over exponentially many feature interactions. This paper introduces Active Circuit Discovery (ACD), a framework that combines attribution-graph analysis with active inference to select interventions efficiently. ACD uses Anthropic’s circuit-tracer library as its attributiongraph backend, applying Edge Attribution Patching with transcoders to identify the active transcoder features for each prompt. A partially observable Markov decision process (POMDP) agent, implemented with pymdp, maintains a multi-factor generative model of feature importance, layer role, and causal influence. At each step, the agent selects both a target feature and an intervention type (ablation, activation patching, or feature steering) by minimising Expected Free Energy over the joint feature–action space, and it learns its observation model online through Dirichlet parameter updates. ACD is an interventionselection layer over existing attribution-graph tools; it is not a whole-circuit discovery method, and no claim of state-of-the-art circuit discovery is made. The framework is evaluated on Gemma-2-2B (26 layers) and Llama-3.2-1B (16 layers) across four settings: Indirect Object Identification (IOI), multi-step reasoning, feature steering, and a multidomain benchmark spanning geography, mathematics, science, logic, and history. With a budget of 20 interventions per prompt, an ablation-only agent scored by bounded oracle efficiency against the ablation oracle reaches 82.0% efficiency on Gemma IOI and 73.0% on Gemma multi-step. It exceeds random selection by 43.5% (relative) on Gemma IOI (paired permutation p = 0.031) and is competitive with greedy ranking, a heuristic UCB bandit, and a plain UCB baseline. A direct Edge-Attribution-Patching ranking is itself a strong baseline that the agent does not consistently surpass, and on Llama multi-step the agent reaches 9.3% efficiency (37.8% with finer layer-role bins). All comparisons report bootstrap 95% confidence intervals. The full multi-action agent is characterised separately by a Relative Cumulative KL, a steering-driven amplification factor reported apart from the bounded efficiency. Feature steering changes the top-1 prediction in a dose-dependent manner, but a matched random-feature control shows that circuit-selected features are only marginally, and not significantly, more steerable than random active features at large multipliers, indicating that part of the effect is generic activation scaling. Multi-domain analysis shows task-dependent circuit structure, with IOI circuits concentrated in late layers and reasoning and scientific knowledge recruiting early and middle layers. Code, notebooks (free T4), AMD64/aarch64 Docker images, and raw results are publicly available. Full article

Existing accounts of platform-mediated complementor growth rest on two limiting assumptions: that platform enablement constitutes a homogeneous environmental input and that firm growth is a unitary outcome. This double simplification obscures how distinct platform provisions generate qualitatively different forms of firm transformation. This study asks which combinations of mechanistically distinct platform enablement types and internal strategic response capabilities activate which forms of leapfrog growth among complementor firms operating under dual institutional governance. We employ fuzzy-set Qualitative Comparative Analysis (fsQCA) on survey data from 374 complementor firms in China’s autonomous driving platform ecosystem. Five antecedent conditions are examined across two dimensions: platform enablement, comprising rule-based enablement (RE) and business platform enablement (BPE); and strategic response capabilities, comprising network linkage capability (NLC), organizational ambidexterity (OA), and policy responsiveness (PR). Three outcome variables capture three non-reducible leapfrog dimensions: technology-chain (TL), value-chain (VL), and institutional-chain (IL) transitions. A reverse-causality robustness check and a common-method-bias assessment corroborate the validity of findings. The analysis identifies equifinal configurational pathways with distinct dominant logics across the three chains. Technology-chain transitions are predominantly network-linkage-driven; value-chain transitions are policy-responsiveness-anchored; institutional-chain transitions exhibit genuine equifinality between network-linkage and policy-responsiveness pathways, both requiring dual-platform enablement as a universal structural precondition. No single enabling condition or capability suffices; leapfrog growth is irreducibly configurational and causally asymmetric. The study offers a dual-enablement, three-chain configurational framework for understanding platform-mediated firm growth under dual institutional governance. For complementor firms, findings support dimension-selective capability investment over uniform accumulation strategies. For platform orchestrators, differentiated governance design calibrated to specific complementor upgrading trajectories outperforms homogeneous resource provisioning. For policymakers, institutionalized consultative channels linking private platform governance with public regulatory processes are recommended to facilitate coordinated digital industrial transformation. Full article

This article argues that the reconstruction of modern knowledge in late Qing China was not merely the result of the passive importation of Western disciplinary categories, nor simply the natural collapse of the traditional Sibu 四部 system. Focusing on Xia Zengyou 夏曾佑 (1863–1924), it shows how Buddhism, especially Mahayana Buddhist reading traditions and classificatory logic, functioned as an indigenous epistemological resource in the reordering of knowledge. Through an analysis of Xia’s personal reading lists and handwritten catalogues, including Shengping Suoxue 生平所學 and his Handwritten Catalogue of Collected Books 手抄藏書書目, this study demonstrates that Xia organized his books in the sequence of Buddhist works, newly translated Western works, and indigenous Chinese texts. This arrangement reversed the Confucian-centered hierarchy of the Sibu system, in which the jing 經 category occupied the privileged position. By comparing Xia’s classificatory practice with those of Shen Zengzhi 沈曾植 (1850–1922), Liang Qichao 梁啟超 (1873–1929), Xu Weize 徐維則 (1867–1919), and Yang Renshan 楊仁山 (1837–1911), the article argues that Xia did not simply adopt Western systems of knowledge. Rather, he used Buddhist textual order, cross-sectarian Mahayana learning, and Buddhist epistemological assumptions to relativize classical authority, accommodate Western learning, and construct a new reading horizon. Buddhism in this case was not only a matter of personal faith or religious revival; it became a conceptual and classificatory tool through which modern knowledge could be made intelligible. The article therefore contributes to the study of religion and modernity by showing that the formation of modern Chinese knowledge was not a purely secular process, but a religiously mediated transformation. Full article

CeO₂ is a representative oxygen-storage oxide because its fluorite lattice can reversibly release and reincorporate oxygen through the Ce⁴⁺/Ce³⁺ redox couple and the associated formation and annihilation of oxygen vacancies. Although doped CeO₂ has been studied extensively, the literature has often treated oxygen-storage enhancement mainly in terms of dopant identity and composition, whereas the more fundamental issue is how a given doping strategy constructs a specific defect state within the fluorite host. Here, oxygen-storage enhancement is discussed from the standpoint of defect-state engineering. The discussion focuses on three routes, as follows: rare-earth single doping, cation–anion co-doping, and route-dependent dopant incorporation. Rare-earth single doping correlates aliovalent substitution with lattice expansion, vacancy generation, and finite oxygen-storage-capacity (OSC) optima. Cation–anion co-doping further shows that simultaneous perturbation of the cationic and anionic sublattices can amplify the defect response, while also demonstrating that vacancy concentration alone does not fully account for OSC enhancement. Route-dependent doping adds an additional dimension by showing that the same dopant can produce different lattice responses, defect populations, and oxygen-release behaviors when introduced through different pathways. On this basis, the review argues that OSC in doped CeO₂ is more meaningfully rationalized through a coupled descriptor set involving lattice accommodation, Ce³⁺/Ce⁴⁺ redistribution, oxygen-vacancy abundance, and dopant incorporation pathway. Taken together, these observations shift the design logic of oxygen-storage ceria from empirical dopant screening toward deliberate defect-state construction. Full article

The regulatory conditions present regarding anthropogenic pollution of two large Chilean Northern Patagonian lakes, Llanquihue and Villarrica, are analyzed. Taking the risk-based approach (RBA) as a reference framework, the study addresses three relevant dimensions of the socio-environmental risks present in the regulatory regimes of each ecosystem: (1) the reflexive components of the institutions involved, (2) the deployment of organizational processes regarding regulatory norms, and (3) the modalities for addressing change and complexity in the regulatory field. Developing a qualitative multiple-case study with criterion-oriented maximum variation sampling, 40 individual interviews conducted with participants who perform tasks in both cases are analyzed, examining their regulatory configurations according to the investigated dimensions. The most important findings account for: (1) an institutional attenuation that bureaucratically minimizes socio-environmental risks, hindering the transition towards preventive approaches marked by a political culture that prioritizes formal compliance over territorial management; (2) a profound institutional fragmentation and centralization of regulation that dilutes responsibility, operating under logics of minimal efforts that prevent the watershed perspective from achieving normative legitimacy; and (3) a regulatory field overwhelmed by wide-ranging phenomena of difficult regulatory management, where the binary classification of saturation proves insufficient to address diffuse pollution and risks of difficult reversibility. It is concluded that strengthening the regulatory capacities of emerging nations regarding the socio-environmental protection of large lakes requires gradually integrating risk as an organizing criterion and prospecting watershed governance based on multiple regimes complementary to the regulatory effort of the already atomized and centralized normative instruments available. Full article

Nature-based solutions (NBSs) involving tree-based interventions deliver multiple community benefits, yet evidence linking these benefits to underlying socio-ecological vulnerabilities remains limited. This study synthesised metadata from 131 European treescape NBS case studies spanning eight biogeographical regions using reverse-logic, thematic qualitative analysis. Case studies were identified via adapted PRISMA guidelines from open-access repositories, with community benefit themes categorised and mapped spatially across bioregions. The analysis revealed eleven principal community benefit categories and distinct region-specific patterns: Mediterranean interventions primarily mitigated extreme heat and drought vulnerabilities, whilst Alpine projects addressed slope stability and hazard reduction. The Continental and Atlantic regions emphasised social cohesion, recreational access, and the preservation of cultural heritage. The reverse-logic methodology successfully identified underlying socio-ecological vulnerabilities through systematic analysis of observed benefit profiles across diverse European contexts. This approach provides evidence-based guidance for designing location-sensitive treescape NBS that advance environmental research and public health objectives. The findings establish a methodological foundation for future assessments of NBS effectiveness and for refining location-specific treescape interventions that address community vulnerabilities and enhance adaptive capacity. Full article

Life Cycle Assessment (LCA) is extensively employed to support sustainability evaluation in waste management and manufacturing systems; however, outcomes are highly sensitive to methodological decisions, particularly end-of-life (EoL) allocation approaches. This study examines how cut-off and substitution approaches affect the energy performance and decarbonisation potential of high-density polyethylene (HDPE) injection moulding systems. A dual framework is adopted: first, a literature review examines methodological sensitivities in EoL modelling; second, a quantitative case study assesses industrial-scale primary data for the production of durable HDPE bottles (300 mL). The LCA model integrates specific technical parameters, including a 220 °C melt temperature and a 36 s cycle time, ensuring a realistic representation of manufacturing conditions. The results indicate that allocation choices significantly influence calculated impacts, sometimes reversing the relative ranking of configurations. Substitution-based approaches report higher benefits by crediting avoided primary production, while cut-off logic provides more conservative estimates. Quantitative analysis shows that transitioning from open-loop to fully closed-loop configurations reduces cumulative energy demand by 3.2% and freshwater emissions per functional unit by 2.8%. Furthermore, the study identifies a ‘landfill paradox’ specific to HDPE waste within transitional energy systems: due to the carbon sequestration effect of landfilled polymers and current grid emission factors, landfilling exhibits a lower net carbon footprint (0.03 kg CO₂-eq./kg) than high-efficiency incineration (1.54 kg CO₂-eq./kg). These findings highlight that circular economy evaluations are strongly shaped by methodological assumptions, with direct implications for energy policy. Bridging the gap between specific industrial processing parameters and end-of-life allocation logic underscores the need to incorporate primary industrial data and transparent allocation frameworks to support reliable decision-making in the transition toward low-carbon and energy-efficient manufacturing systems. Full article

AI infrastructure is entering a constraint-dominated regime in which power access, cooling, water conditions, reliability, and financing jointly shape cost, sustainability, and operational risk. Yet the metrics used to evaluate these systems remain fragmented across facility engineering, compute/workload performance, and economic or risk analysis, with definitions that often sit at different layers and under different boundaries. This fragmentation weakens cross-layer reasoning and makes decision-traceable trade-off analysis difficult. This paper proposes a structured, decision-oriented measurement architecture for AI infrastructure metrics. The framework combines a 6 × 3 taxonomy, which organizes metrics across six layers and three semantic domains, with a procedural workflow built around a problem card, variable registry, minimality gate record, activated-cell map, boundary log, metric ledger, and a results sheet with case-pack manifest. Within this protocol, the Metric Propagation Graph is used as a case-specific dependency representation for tracing decision-facing metrics back to minimal boundary-consistent inputs. It is introduced as a traceability layer within the framework rather than as a stand-alone graph-theoretic method. The paper is illustrated through one fully worked case and one scoped portability illustration. The first is a fully worked large-load planning case for the Northern Virginia data-center corridor within PJM’s Dominion zone, showing that a boundary-consistent integrated metric can reverse the ranking obtained under a simpler screening view. The second is a scoped portability illustration for hourly matching under dual Scope 2 boundaries. Its purpose is not to provide a second full empirical validation, but to show how the same dossier logic, boundary discipline, and traceable metric construction transfer to a distinct decision setting. Full article

Intermittently operated, tankless reverse osmosis (RO) systems are widely used in decentralized and point-of-use applications, yet water stagnation during idle periods remains a critical challenge, leading to degraded water quality, accelerated fouling, and performance loss. This study presents an experimentally validated engineering solution that eliminates stagnant water in intermittently operated RO systems. A dual-membrane RO configuration with flexible series–parallel switching was developed, enabling membranes to alternate between production and flushing modes. An adaptive control strategy, integrated into the system hardware, regulates membrane switching and flushing based on real-time feed-water quality. The proposed configuration and control framework was evaluated under representative intermittent operating conditions. Experimental results show that the zero-stagnant-water strategy effectively prevents residual water accumulation during shutdown and maintains stable permeate quality, with total dissolved solids consistently below 10 mg/L. Long-term testing further demonstrates reduced membrane fouling and slower performance degradation compared with conventional fixed-operation schemes, resulting in enhanced desalination efficiency and operational stability. Owing to its modular design and simple control logic, the proposed approach is readily transferable to decentralized and point-of-use membrane water treatment systems requiring reliable, high-quality water under intermittent operation. Full article

To address the poor adaptability and rigid initiation modes of the loitering munition fuze in complex environments and the inadequacy of single fuzzy control against strong interference, this paper proposes a dual-loop adaptive reconfiguration control method. The architecture integrates the Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm with fuzzy logic. The inner loop uses TD3 to dynamically optimize fuzzy scaling factors based on real-time interference and state deviations. Concurrently, the outer loop utilizes a Fuze Readiness Index (FRI) and a finite state machine to manage real-time multi-modal mission switching (e.g., proximity, delay, and airburst) and reverse safety-state conversions. Co-simulations under non-stationary composite interference show that the proposed method reduces the burst height RMSE by 82.4% and 61.6% compared with the fixed-threshold and standard fuzzy baselines under the considered non-stationary composite interference setting, respectively. The false alarm rate (FAR) is reduced to 0.15%, and the reconfiguration response time under sudden interference is shortened to 12 ms. Even under extreme conditions, such as 400 ms sensor signal loss, the relative error remains within 5%. These simulation results demonstrate the potential of the proposed architecture to improve precision, responsiveness, and robustness under dynamic interference conditions and show good robustness to intermittent observation loss within the simulated operating envelope. Full article

The latest H266/VVC standard has received numerous praises for its excellent compression efficiency. However, its extremely high computational complexity has become a hindrance to the VVC adaptation industry ecosystem, while also increasing the difficulty of hardware design and application costs. To address this issue, we designed an efficient intra-coding scheme based on neural networks, which consists of three parts: Firstly, we designed a neural network-based reverse prediction algorithm that uniquely utilizes the CNN’s prediction results for lower-level blocks to determine the QTMT partitioning of upper-level blocks, cleverly solving the adaptation problem of existing models to complex VVC partitioning patterns—a decision-making logic that has not been fully explored. Secondly, we designed a pruning algorithm, which is the first to dynamically couple the real-time RDO cost of BT segmentation with the TT segmentation direction, achieving adaptive decision-making. Finally, we designed a complexity pre-screening module. On the basis of analyzing whether the CU texture is smooth, this module designs four sets of adaptive thresholds for non-square CUs introduced in VVC. These thresholds can dynamically adjust local and global thresholds based on CU size, enabling size sensitive texture evaluation to determine whether the current block needs further partitioning. The experimental results show that, compared with traditional VTM4.0, our method reduces the average encoding time by 49.21%, while the BD-BR increase is 1.61%, and the BD-PSNR decreases by 0.06 dB, fully demonstrating its superiority and performance balance. Full article

The proliferation of intelligent edge devices demands compact, low-power hardware capable of dynamically switching between sensing, logic, and learning tasks—a versatility that traditional multi-chip solutions fundamentally lack. Here, we demonstrate a reconfigurable spin–orbit torque (SOT) device based on an FeTb/Ru/Co synthetic antiferromagnetic (SAF) heterostructure. By modulating the input current amplitude, the device dynamically switches between two distinct operating modes: saturation and activation. In the saturation regime (>80 mA), deterministic magnetization reversal enables Boolean logic operations (AND, NOR). In the activation regime (<80 mA), gradual, non-volatile conductance modulation emulates synaptic plasticity. Benefiting from the strong antiferromagnetic coupling and near-zero net magnetization of the SAF structure, all operations are achieved without external magnetic fields. This single-device, dual-mode reconfigurable architecture establishes a new paradigm for high-density, low-power, multifunctional in-memory computing units, with promise for advancing adaptive edge computing chips. Full article

Global urbanization has long been hampered by the “metrocentric priority” paradigm, with small towns—core hubs for urban–rural integration—severely undervalued in practical value. Amid China’s transition to high-quality urban–rural integration, unbalanced small town development has become a critical bottleneck for county-level factor flows, demanding systematic research to unlock their strategic value and resolve urban–rural dual predicaments. Existing studies suffer from scientific gaps including unidirectional linear cognition, insufficient complex system thinking, and weak interpretation of regional heterogeneity, remaining at the stage of static correlation description and failing to reveal the two-way reciprocal feedback logic between small towns and urban–rural integration. Meanwhile, the application of complex system theory in urban–rural research is still confined to theoretical narratives, which hinders the advancement of research from descriptive analysis to mechanism interpretation. Taking Henan Province (a typical agricultural and populous province reflecting China’s urban–rural development) as a case, this study builds a “local emergence–global synergy” framework based on complex system theory, establishes a dual indicator system for small towns’ multidimensional performance and county-level urban–rural integration, and integrates spatial statistical analysis, bidirectional regression and coupling coordination models to explore their cross-scale spatiotemporal evolution and reciprocal feedback during 2019–2023. Findings show the following: (1) The multidimensional performance of small towns presents a pattern characterized by polarized expansion of high-value regions and overall improvement of low-value regions, while county-level urban–rural integration evolves into a polycentric structure featured by central dominance and southern growth. (2) There is a significant two-way asymmetric relationship between small towns’ multidimensional performance and county-level urban–rural integration: the positive effect is significantly stronger than the reverse effect, and both direct impacts are significantly weakened after introducing economic variables, indicating that economic development serves as a key transmission channel. (3) The coupling mechanism presents three evolutionary paths with pronounced core–periphery spatial heterogeneity. Grounded in complex system theory, this study constructs a systemic analytical framework of “local emergence of small-town subsystems and global synergy of county-level systems”, verifies the core proposition of two-way interactions between subsystems and the overall system in the urban–rural complex giant system, and enriches the localized application of complex system theory and the urban–rural continuum theory in traditional agricultural regions of China. This study provides a foundational empirical paradigm for the in-depth exploration of nonlinear characteristics and threshold effects in future research. It offers theoretical support for policy formulation of county-level urban–rural integration in traditional agricultural regions of China, and it provides Chinese experiences for the Global South with similar contexts to explore inclusive urbanization pathways, promoting cross-cultural dialogue and practical transformation of urban–rural integration theory. Full article

Can machine learning generate statistically validated alpha in equity markets while adapting to changing market conditions? This study addresses this question by proposing a regime-aware LightGBM framework conditioned on market regimes detected via a rolling Hidden Markov Model, eliminating look-ahead bias. Backtested on 51 NASDAQ-100 constituents (2015–2026), the strategy achieved a portfolio Sharpe ratio of 1.18 (95% CI: [0.53, 1.84]) and outperformed four baseline models. The key findings include the following: (i) cross-asset features (Bitcoin as a leading indicator) contribute the most predictive value; (ii) macroeconomic indicators outweigh traditional technical indicators for high-beta stocks; (iii) the model autonomously adapts its decision logic across regimes, shifting from mean reversion in bear markets to risk appetite monitoring in bull markets. While block bootstrap tests confirm statistical significance ($p<0.001$), the Deflated Sharpe Ratio (0.69) does not reach formal significance after multiple testing correction—an honest finding we report transparently. Full article