Organoids, which are three-dimensional (3D) multicellular structures derived from stem cells or tissue-specific progenitors, have emerged as a transformative platform for drug evaluation within tissue engineering and regenerative medicine (TERM). These models recapitulate human tissue complexity with greater fidelity than traditional two-dimensional cultures and animal models do, offering significant advantages in predicting human-specific drug responses, enabling personalized disease modeling, and accelerating drug development. This review critically examines advances, strategies, and challenges associated with the application of organoids for drug testing in TERM. We discuss diverse organoid types, including hepatic, cardiac, neural, gastrointestinal, lung, and tumor models, and their specific applications in assessing organ-specific toxicity, drug metabolism, and multiorgan interactions. Innovative methodologies such as organ-on-a-chip integration, multiorgan systems, and 3D bioprinting are highlighted as pivotal strategies for enhancing the physiological relevance and scalability of organoid models. Despite their considerable promise, organoids present several challenges, including limitations in reproducibility, long-term culture maturity, and functional complexity. Furthermore, ethical and regulatory considerations, particularly concerning patient-derived models and genetic modifications, must be addressed to facilitate the clinical translation of organoid-based drug testing. Finally, we explore future directions, including the integration of artificial intelligence-driven predictive models, clustered regularly interspaced short palindromic repeats (CRISPR)-based genome editing, and vascularization strategies, which hold potential for overcoming existing limitations and advancing the field of drug evaluation in regenerative medicine.
Organoids have emerged as powerful tools for disease modeling, mechanistic exploration, drug screening, and regenerative medicine. Spinal cord and peripheral nerve organoids mimic real organs in terms of cell morphology, tissue structure, and function, offering new possibilities for tissue engineering therapies in spinal cord injury (SCI) and peripheral nerve injury (PNI). In this review, we introduce the genesis and structure of the spinal cord and peripheral nervous system and summarize the key factors for constructing spinal cord and peripheral nerve organoids. We then discuss the recent advances in the formation of three-dimensional (3D) organoid structures through engineering approaches and explore their potential applications in neural tissue engineering. Finally, we address current challenges such as the high heterogeneity among neural organoids and incomplete replication of organ functions. Future development of engineered matrix materials and manufacturing methods is crucial for further clinical translation of organoids in spinal cord and peripheral nerve repair.
Peripheral and central nerve injuries present significant challenges in clinical practice due to the paucity of effective therapeutic options. Extracellular vesicles (EVs), lipid membrane-encapsulated nanoparticles released from most cell types, are enriched in bioactive molecules capable of mediating intercellular communication. Recently, EVs have shown potential for applications in regenerative medicine, notably as alternatives to cell therapy. A therapeutic paradigm shift is emerging towards the EV-based treatment of nerve injury. Accumulating evidence indicates that EVs promote nerve regeneration and functional recovery, primarily by modulating neural cell function, immune-inflammatory responses, and angiogenesis. Various strategies have been developed to incorporate EVs into nerve grafts or to utilize them as diagnostic and therapeutic vehicles. This review aimed to provide a systematic survey of the available literature on the latest progress in EV-based treatments for peripheral nerve, spinal cord, and brain injuries. Additionally, prospects of EV-based nerve injury repair are discussed in terms of deep mechanistic understanding, technological innovations, and interdisciplinary convergence.
This study established human cerebral organoids as a promising platform for investigating central nervous system oxygen toxicity (CNS-OT). Through integrated transcriptomic, functional, and pharmacological analyses, we demonstrate that hyperbaric oxygen (HBO) exposure triggers pressure-dependent neurotoxicity mediated by the reactive oxygen species (ROS)-lysosome-mechanistic target of rapamycin (mTOR) axis. Key findings include the following: mechanistic hierarchy: Five atmospheres absolute (ATA) HBO induces metabolic dysregulation and cell cycle arrest, whereas six ATA exceeds compensatory thresholds, triggering overt apoptotic signatures; pathway crosstalk: Lysosomal permeabilization activates mTOR complex 1 (mTORC1) via cathepsin release, while mTORC1 hyperactivation suppresses transcription factor EB (TFEB)-mediated lysosomal regeneration, creating a self-amplifying loop; therapeutic potential: Mouse validation confirmed that mTOR inhibition (temsirolimus) attenuates neurotoxicity, with hippocampus-specific efficacy. The cerebral organoid model offers a human-relevant system to overcome species limitations in neurotoxicity research, facilitating mechanistic discovery and therapeutic target identification.
The subcutaneous space is an advantageous site for cell transplantation because of its minimally invasive accessibility. However, poor vascularization often compromises cell survival, leading to significant apoptosis of transplanted cells. To address this limitation, we engineered a pre-vascularized niche by subcutaneously implanting polyvinyl chloride (PVC) catheters into rodent models. Foreign body reactions elicited by PVC catheters recruited reparative immune cells, which secreted pro-angiogenic factors to induce robust neovascularization. The subsequent transplantation of stem cell-derived vascular organoids into this engineered niche resulted in optimal engraftment, forming functional, perfused human blood vessels. In a proof-of-concept islet transplantation model, even a marginal dose of islets rapidly restored normoglycemia, which was attributed to enhanced graft revascularization. Collectively, our findings demonstrate a simple yet effective strategy for overcoming the vascular limitations of subcutaneous cell transplantation, offering promising potential for clinical applications.
Patient-derived organoids (PDOs) are valuable for predicting the anticancer efficacy of therapeutics. However, conventional Matrigel-based PDO culture systems impede cellular therapy testing by preventing direct contact between cell therapies and organoids. In this study, we present a novel microwell array for culturing Matrigel-free PDOs to evaluate the in vitro anticancer activity of chimeric antigen receptor T (CAR-T) cells. We fabricated the microwells with a multi-layer nested architecture using a molding process, in which the casting molds were precision-customized via rapid three-dimensional (3D) printing. This microwell system enables the cultivation of organoids without Matrigel. Using this platform, we successfully cultured Matrigel-free lung cancer organoids (LCOs) derived from lung adenocarcinoma patients harboring epidermal growth factor receptor (EGFR) mutations. Similar to primary cancer cells, the cultured LCOs exhibited high EGFR protein expression. EGFR-CAR-T cells efficiently targeted and lysed Matrigel-free LCOs in the microwells but failed to kill conventionally cultured LCOs grown in Matrigel. Furthermore, EGFR-CAR-T cells effectively inhibited the growth of PDO xenograft (PDOX) tumors in vivo, consistent with the in vitro LCO results. These findings demonstrate that microwell-cultured, Matrigel-free LCOs provide an in vitro platform for assessing CAR-T cell efficacy, paving the way for standardized organoid models in cell therapy development.
Interorgan interactions are essential for organogenesis and maturation, with their dysregulation leading to developmental disorders. However, the ability of current physiologically relevant human models to recapitulate interorgan crosstalk during early developmental stages remains limited. Here, we develop a trans-germ-layer codevelopment organoid chip (TGCO-Chip) that enables the coemergence of two interconnected distinct organoids from a common upstream-lineage stem cell aggregate under well-controlled biochemical conditions. Specifically, we established a human pluripotent stem cell-derived heart-brain codevelopoid (HBC) model using a TGCO-Chip, and the codevelopoid recapitulated the developmental features of the heart and brain, including cell lineages, tissue architecture, and functionality. Furthermore, codevelopoids emulate neural projections to cardiac tissues and their regulatory effects during the early developmental stage of organogenesis. Compared with the interconnected heart-heart organoids, the neural compartment significantly increased the average cardiac beating rates and contraction amplitudes. Transcriptomic analysis confirmed that neural compartments in HBCs promoted cardiac differentiation and maturation. Overall, the TGCO-Chip platform provides an innovative tool for bioengineering multiorganoid complexes derived from shared progenitor lineages. Codevelopoids hold immense potential for applications in developmental biology, disease modeling, and regenerative medicine and can provide unprecedented insights into the dynamic interactions between different cell lineages and tissues.
Ovarian cancer remains a highly lethal gynecologic malignancy. Early diagnosis poses significant challenges, and the five-year survival rate is consistently below 45%. Current standard-of-care combines surgical resection with platinum-based chemotherapy. Emerging therapeutic modalities like chimeric antigen receptor-T (CAR-T) therapy show promise, though they face efficacy constraints due to tumor heterogeneity and immunosuppressive microenvironments. Conventional models including two-dimensional (2D) cultures and patient-derived xenografts are increasingly supplanted by organoid and tumor-on-a-chip technologies due to intrinsic limitations and poor clinical translatability. This study established multiple tumor-on-a-chip platforms derived from ovarian cancer organoids and conducted systematic in vitro drug sensitivity screening. Furthermore, by utilizing patient-derived organoids to engineer multicellular dynamic microenvironments, we achieved one of the extremely limited evaluations of CAR-T efficacy against solid tumors within ovarian cancer microfluidic systems. This work establishes an enhanced preclinical platform to advance therapeutic screening and personalized treatment development.
Lung cancer, as one of the leading causes of cancer-related deaths worldwide, exhibits complex pathogenesis, with the association between inflammation and malignant transformation drawing significant attention. This paper focuses on the pivotal role of phospholipid metabolism in the inflammation-to-cancer transition of lung cancer. It systematically elucidates the molecular mechanisms by which phospholipid metabolism drives this transition, its impact on the immune microenvironment, its involvement in cell death resistance processes, and clinical translation strategies from lung cancer to pan-cancer types. Furthermore, it explores controversies and future prospects in phospholipid metabolism research. Through comprehensive analysis of relevant literature, this review aims to provide novel insights and theoretical foundations for the prevention, diagnosis, and treatment of lung cancer.
This article seeks to explore a general formulation of artificial general intelligence (AGI) under a unified framework that defines AGI agents as points in a joint (C, U, V) space. An agent is characterized by three components: ① a cognitive architecture C, which represents the modules (mathematical functions) inside the agent’s mind, as well as the connections and communication protocols between these modules, including the theory of mind (ToM); ② a set of potential functions U, which represents the skills of perception, cognition, and planning (e.g., a potential function can be a neural network trained for visual object recognition, or embodied motion planning); and ③ a set of value functions V, which includes the agent’s urges, preferences, and social affections, as well as benefits for individual agents or a group of agents. In this setting, “intelligence” is defined as a wide range of phenomena exhibited by agents when they interact with complex environments (i.e., physical intelligence) and other agents (i.e., social intelligence). Given an initial point in the (C, U, V) space, an agent can explore new V-dimensions, which in turn drives the acquisition and learning of skills by enabling the learning of new potential functions U in the environment and by updating the cognitive model. We have developed a Tong test as a benchmark and evaluation criteria: An agent that has reached the human level (C, U, V) is called a “Tong Agent.” The convergence of this process defines the limits of the agent’s evolution; we name this the “stopping problem” of Tong Agents, based on the analogy of the halting problem in a Turing machine.
The advent of foundation models (FMs), large-scale pre-trained models with strong generalization capabilities, has opened new frontiers for financial engineering. While general-purpose FMs such as GPT-4 and Gemini have demonstrated promising performance in tasks ranging from financial report summarization to sentiment-aware forecasting, many financial applications remain constrained by unique domain requirements such as multimodal reasoning, regulatory compliance, and data privacy. These challenges have spurred the emergence of financial foundation models (FFMs): a new class of models explicitly designed for finance. This survey presents a comprehensive overview of FFMs, with a taxonomy spanning three key modalities: financial language foundation models (FinLFMs), financial time-series foundation models (FinTSFMs), and financial visual-language foundation models (FinVLFMs). We review their architectures, training methodologies, datasets, and real-world applications. Furthermore, we identify critical challenges in data availability, algorithmic scalability, and infrastructure constraints and offer insights into future research opportunities. We hope this survey can serve as both a comprehensive reference for understanding FFMs and a practical roadmap for future innovation.
Vision-language segmentation models (VLSMs) are effective in medical image segmentation tasks. However, a major limitation of these models is their dependence on manually crafted textual inputs. Studies have used visual question answering to semiautomatically generate textual information. However, these methods encounter challenges such as error accumulation. Herein, we propose a method to learn conceptual text prompts directly from visual regions of interest (ROIs) for facilitating medical image segmentation. We extracted textual conceptual attributes from ROIs using a large multimodal model to derive coarse real-text prompts. A text latent space transformation module accepted the ROI images as input for generating fine-grained pseudo-text prompts to compensate for the lack of image detail perception in the abovementioned real-text prompts. These prompts were encoded into a unified text embedding. Thereafter, we applied a self-adding noise knowledge distillation method to transfer the knowledge from text embedding to the class token of the image encoder, enabling direct text-guided inference during testing while reducing error accumulation. Our approach minimized the need for manual prompt design by leveraging explicit discrete and implicit continuous text prompts to effectively guide visual segmentation. Extensive evaluation across 13 medical image segmentation datasets demonstrated that our model outperformed the state-of-the-art VLSMs and vision-based segmentation models, exhibiting superior segmentation accuracy.
The automatic synthesis of analog circuits presents significant challenges. Most existing approaches formulate the problem as a single-objective optimization task, overlooking the fact that design specifications for a given circuit type can vary widely across applications. To address this limitation, we introduce specification-conditioned analog circuit generation, a task that directly generates analog circuits based on stated specifications. The motivation is to find an effective method that leverages existing well-designed circuits to improve automation in analog circuit design. Specifically, we propose CktGen, a simple yet effective variational autoencoder model that maps discretized specifications and circuits into a joint latent space and reconstructs the circuit from that latent vector. Notably, as a single specification may correspond to multiple valid circuits, naively fusing the specification information into a generative model does not capture these one-to-many relationships. To address this, we first decouple the encoding process of circuits and specifications and align their mapped joint latent space. Then, we employ contrastive training with a filter mask to maximize differences between encoded circuits and specifications. Furthermore, classifier guidance along with latent feature alignment promotes the clustering of circuits sharing the same specification, thus avoiding model collapse into trivial one-to-one mappings. By canonicalizing the latent space with respect to the specifications, we can further optimize and search for an optimal circuit that satisfies the valid target specification. We conduct comprehensive experiments on the open circuit benchmark and introduce several metrics to evaluate cross-model consistency in the specification-conditioned circuit generation task. The experimental results demonstrate that CktGen achieves substantial improvements over existing state-of-the-art methods.
In smart manufacturing, autonomous mobile robots play an indispensable role in conducting inspection and material handling operations, yet they face significant limitations regarding adaptability and resilience within unstructured environments. Vision and language navigation (VLN), a human-guided navigation paradigm, emerges as a compelling solution to these challenges. Nevertheless, VLN’s practical implementation is constrained by limited task generalization capabilities, inadequate response to diverse linguistic commands, and insufficient consideration of sensor-induced noise in environmental perception. This research addresses these limitations by introducing an innovative vision-language model (VLM)-based human-guided mobile robot navigation approach in an unstructured environment for human-centric smart manufacturing (HSM). This approach encompasses three-dimensional (3D) robust scene reconstruction through advanced point cloud techniques, zero-shot semantic segmentation via a VLM, and natural language processing through a large language model (LLM) to interpret instructions and generate control code for navigation. The system’s efficacy is validated through extensive experiments in an unstructured manufacturing setup.
With the rapid deployment of lithium iron phosphate (LFP) batteries and their finite service life, the annual accumulation of end-of-life LFP batteries has risen substantially. This growing accumulation creates a range of safety and environmental concerns, including leakage, thermal runaway, combustion, and explosion, which threaten natural environments such as water, soil, and air, while also endangering both human and wildlife safety. Therefore, the effective and responsible recycling of spent LFP batteries is crucial. Recycling not only serves as a key approach to converting waste streams into valuable resources but also mitigates the relevant environmental concerns. The recovery of valuable components, particularly lithium, supports resource sustainability and provides environmental, economic, and societal benefits. Among the components of spent LFP batteries, lithium is the most valuable, primarily because these batteries generally have a lower intrinsic recycling value than other lithium-ion batteries (LIBs) and do not contain economically high-value metals such as nickel and cobalt. However, the current industrial recovery rate of lithium from spent LFP batteries remains below 1%, underscoring the urgent need for further development of efficient lithium recovery technologies. Selective lithium leaching has emerged as a highly attractive and environmentally benign approach tailored for lithium recycling, receiving growing attention from both academia and industry. Various selective leaching techniques have been developed, including chemical selective leaching, electrochemical selective leaching, bio-selective leaching, leaching-precipitation, and direct selective leaching, all designed to selectively recover lithium from spent LFP batteries. Despite differences in operational approaches, these methods are founded on comparable thermodynamic principles and recovery goals. This review systematically summarizes recent technological developments and research progress, and integrates thermodynamic potential (E)-pH diagram analysis to evaluate the feasibility, advantages, and limitations of various selective leaching methods. Economic feasibility, operational complexity, and environmental performance are systematically evaluated. Furthermore, the key characteristics, limitations, and practical applicability of these technologies are comparatively discussed, providing a systematic comparison, critical assessment, and prioritization of all current research strategies in terms of industrial feasibility and future development potential. Additionally, this review highlights eight major advantages and five potential development directions for selective lithium leaching, emphasizing its promising role in future lithium recycling systems. Finally, based on selective leaching strategies, a comprehensive process flowchart for the overall recycling of LFP batteries is proposed as a conceptual framework for future industrial implementation.
The application of artificial neural network (ANN) models to achieve higher accuracy in industrial sensing has become a popular research topic in recent years. However, neural network models are purely data-driven multivariate “black-box” models, and the features extracted from the hidden layer have no actual physical meaning, making the performance of ANN-based sensing models unstable and difficult to practically apply at process industry sites. To address these challenges, this paper proposes a generalized ANN model called the partial least squares (PLS)-assisted optimization network (PLSaoNET). PLSaoNET employs the PLS model to assist in determining the initialization weights of the network and the number of hidden-layer neurons. The subsequent training serves as a reoptimization process guided by the PLS regression result, enabling the network to incorporate statistical constraints and thereby reducing its reliance on data. In addition, to address the problem of uneven distributions of sample labels at industrial sites, this paper designs a stratified sampling method for network retraining. The efficiency and superiority of the proposed method are verified via two industrial sensing applications: the monitoring of iron grade in iron ore concentrate slurry samples based on laser-induced breakdown spectroscopy (LIBS) data, and the assessment of the quality of diesel fuels based on near-infrared (NIR) spectroscopy data. In comparison with a PLS regression model and a Xavier initialization-based backpropagation neural network (BPNN) model, PLSaoNET exhibits the best modeling accuracy and generalization performance. This work designs a complete theoretical framework to guide the determination of hyperparameters and specify the solution paths of the network, thereby satisfying the triple requirements of accuracy, robustness, and ease of use in industrial processes. The proposed model holds great potential for improving the accuracy and reliability of industrial sensing in production processes.
The decarbonization of China’s steel sector illustrates a central paradox of industrial transformation: Technologies that can deliver deep emissions reductions remain constrained by resource availability, deployment feasibility, and regional disparities. Drawing on the Multi-resolution Emission Inventory for China (MEIC) 2010-2023, this perspective situates the challenge against an empirical baseline where national CO2 totals rose from 8.2 Gt (2010) to 11.2 Gt (2023), with industry and power emissions are tightly coupled, and a handful of regions—Hebei, Shandong, Jiangsu, Inner Mongolia, and Guangdong—exerting disproportionate influence. Within this context, carbon capture, utilization, and storage (CCUS) and hydrogen-based direct reduced iron (H2-DRI) emerge as the two most prominent pathways, yet both present significant limitations often obscured by macro-level comparisons. CCUS offers the largest near-term abatement through retrofits to existing blast furnace-basic oxygen furnace assets, with potential contributions exceeding 40% of industry reductions by 2060. However, full-chain accounting reveals high energy penalties and concentrated water burdens, raising concerns over long-term sustainability. H2-DRI, by contrast, achieves near-zero process emissions under moderate renewable hydrogen supply but faces diminishing returns at aggressive deployment levels, where reliance on grid electricity and fossil-derived hydrogen erodes life-cycle benefits—indeed, emission intensities increase more than six-fold when renewable supply saturates. Economic comparisons are equally boundary-sensitive: CCUS costs hinge on capture and storage integration, while H2-DRI depends on electricity pricing, electrolyzer utilization, and hydrogen transport infrastructure—factors often excluded in optimistic projections. A viable transition therefore requires more than technological substitution. Demand reduction, material efficiency, and scrap recycling must complement region-differentiated strategies, while disruptive innovations in hydrogen transport, electrolytic ironmaking, and capture efficiency will be essential. The steel industry’s trajectory thus becomes a decisive test case for whether large-scale industrial decarbonization can succeed under the real-world constraints of resource scarcity, economic feasibility, and governance capacity.
Current decarbonization efforts are falling short of meeting the net-zero greenhouse gas (GHG) emission target, highlighting the need for substantial carbon dioxide removal methods such as direct air capture (DAC). However, integrating DACs poses challenges due to their enormous power consumption. This study assesses the commercial operation of various DAC technologies that earn revenue using monetized carbon incentives while purchasing electricity from wholesale power markets. We model four commercial DAC technologies and examine their operation in three representative locations including California, Texas, and New York in the United States. Our findings reveal that commercial DAC operations can take financial advantage of the volatile power market to operate only during low-price periods strategically, offering a pathway to facilitate a cost-efficient decarbonization transition. The ambient operational environment such as temperature and relative humidity has non-trivial impact on abatement capacity. Profit-driven decisions introduce climate-economic trade-offs that might decrease the capacity factor of DAC and reduce total CO2 removal. These implications extend throughout the entire lifecycle of DAC developments and influence power systems and policies related to full-scale DAC implementation. Our study shows that DAC technologies with shorter cycle spans and higher flexibility can better exploit the electricity price volatility, while power markets demonstrate persistent low-price windows that often synergize with low grid emission periods, like during the solar “duck curve” in California. An optimal incentive design exists for profit-driven operations while carbon-tax policy in electricity pricing is counterproductive for DAC systems.
Biohydrogen, produced via microbial fermentation of biomass waste, is poised to play a pivotal role in China's green energy transition. Nonetheless, significant obstacles such as high costs, unstable production dynamics, regulatory and metabolic inefficiencies, and limited actual hydrogen yields hinder large-scale application. Addressing these challenges necessitates the integration of machine learning and synthetic biology, forming a robust pathway to enhanced process efficacy and output consistency. The convergence of artificial intelligence (AI) and biotechnology (BT) is revolutionizing biohydrogen production by shifting from traditional empirical methodologies to predictive, engineering-based frameworks. AI equips researchers to interpret and optimize complex metabolic and genetic networks through machine learning and genome-scale modeling. Concurrently, BT is evolving to manipulate microbial communities holistically via synthetic ecology and dynamic modeling. Here, we propose a “digital microbial community” paradigm, intergating multi-scale metabolic modeling and emergent property prediction, AI-powered ecological niche decomposition and closed-loop BT enhanced evolutionary framework for continuous optimization of digital twins through experimental feedback. This fusion facilitates the rational design and real-time optimization of programmable microbial ecosystems, greatly enhancing biohydrogen producing control and efficiency. The transition to digital and data-driven design, utilizing multi-omics and ecosystem-level analytics, further bolsters precision and scalability. While moving from single cells to complex microbial consortia introduces challenges, such as non-linear dynamics and ecosystem stability, the synergy of AI and BT underpins the intelligent, resilient, and sustainable production of biohydrogen, thereby reinforcing its potential as a foundational component of China's renewable energy landscape.
Sulfur-based denitrification (SADeN) technology is a cost-saving and low-carbon alternative for treating organic-deficient wastewater. However, the use of this technology in full-scale applications is still limited. Crucial concerns for biological processes, such as changes in seasonal performance and the ways in which it responds to low temperatures, have yet to be studied in real SADeN processes. Herein, two SADeN biofilters in parallel (downflow, with a design water treatment capacity of 8000 m3·d−1 for each) were systematically investigated over a period of more than 400 days. Seasonal variations in denitrification rates were observed, with an average difference of up to 2 times between summer and winter. The use of thiosulfate was verified as an efficient strategy to improve the performance of SADeN biofilters in winter, which was found to have an overstoichiometric enhancement effect (OSEE), with 24.79%-331.50% more nitrate removal than that calculated according to thiosulfate dosages. Analyzing the biofilter vertically revealed that the OSEE occurred because thiosulfate rapidly consumed dissolved oxygen in the upper zone of the bed and activated the electron flux of the sulfur-based reactive filler (SReF) in the lower zone. Microbial community analysis further suggested that this increase in electron flux may be associated with the abundance recovery of sulfur autotrophic denitrifiers and the stability recovery of the microbial eco-networks. This study offers a paradigm for the full-scale application of SADeN technology in real wastewater treatment plants, providing an in-depth understanding of the role of dosing thiosulfate in tackling low-temperature challenges.
Crop diseases represent a significant threat to global agricultural productivity and food security. The advancement of non-invasive and efficient crop health monitoring technologies is critical for sustainable crop protection and yield stability. The continuous progress of sensing systems and computational methodologies offers promising avenues for developing intelligent agricultural disease monitoring systems. This review systematically evaluates existing research from three dimensions: sensors and systems, methods and algorithms, and applications. It provides an in-depth analysis of the roles of different sensors and systems, discusses key methods and techniques, prediction, and early warning, and explores their applications in real-world agricultural scenarios. Furthermore, this paper identifies the main challenges in agricultural disease surveillance research, particularly in the development of real-time detection techniques, the construction of early-warning models, and the promotion of data sharing and collaboration. Finally, innovative directions and application prospects are explored for integrating crop disease monitoring with big data, artificial intelligence (AI), and the Internet of Things (IoT). These research advances are expected to open new avenues for theoretical innovation and practical applications in crop disease monitoring.
Algae oil contains polyunsaturated fatty acids (PUFAs) that are beneficial to human health and development. Enzymatic refining is a sustainable approach that can improve the quality and nutritional value of algae oil. However, this approach involves the use of specific enzymes to remove unwanted impurities such as phospholipids (PLs) and free fatty acids (FFAs). This study proposes a novel immobilized phospholipase, immobilized phospholipase A1 (PLA1@MCM-41-C8), which serves a dual purpose: it is effective for both degumming arachidonic acid (ARA)-enriched algal oil and its subsequent deacidification. The use of PLA1@MCM-41-C8 achieved a degumming rate of 95.9% for crude algal oil, with an ARA retention rate of up to 97.3%. The acidity of the degummed oil decreased from 14.5 to 1.1 mg KOH·g−1, while simultaneously producing 29.2% high-value diacylglycerol (DAG). By comparing enzymatically degummed and deacidified ARA-enriched algae oil (EDDO) with crude ARA-enriched algae oil (CO), commercialized refining algae oil (CRO), it was found that enzymatic refining had minimal impact on antioxidant stability, fatty acid composition, and oil content. In particular, the flavor characteristics of algal oil before and after enzymatic refining with PLA1@MCM-41-C8 remained unchanged and were dominated by the main volatile compounds: alcohols and aldehydes. Compared with free phospholipase A1 (PLA1), PLA1@MCM-41-C8 exhibited significantly improved thermostability, pH tolerance, solvent tolerance, and long-term reusability. These characteristics make it a promising candidate for complex oil refining in industrial applications.