Global agricultural systems face unprecedented challenges from climate change, resource scarcity, and rising food demand, requiring transformative solutions. Artificial intelligence (AI), particularly deep learning (DL), has emerged as a critical tool for agricultural monitoring, yet a systematic synthesis of its applications remains understudied. This paper presents a comprehensive bibliometric and knowledge graph analysis of 650 + publications (2000–2024) to map AI’s role in agricultural information identification, with emphasis on DL and remote sensing integration (e.g., UAVs, satellites). Results highlight Convolutional Neural Networks (CNNs) as the dominant technology for real-time crop monitoring but reveal three persistent barriers: (1) scarcity of annotated datasets, (2) poor model generalization across environments, and (3) challenges in fusing multi-source data. Crucially, interdisciplinary collaboration—though vital for scalability—is identified as an underdeveloped research frontier. It is concluded that while AI can revolutionize agriculture, its potential hinges on improving data quality, developing environment-adaptive models, and fostering cross-domain partnerships. This study provides a strategic framework to accelerate AI’s integration into global agricultural systems, addressing both technical gaps and policy needs for future food security.

Recent advances in graph neural networks (GNNs) have enhanced multimodal recommendation systems’ ability to process complex user–item interactions. However, current approaches face two key limitations: they rely on static similarity metrics for product relationship graphs and they struggle to effectively fuse information across modalities. We propose MR-CSAF, a novel multimodal recommendation algorithm using cross-self-attention fusion. Building on FREEDOM, our approach introduces an adaptive modality selector that dynamically weights each modality’s contribution to product similarity, enabling more accurate product relationship graphs and optimized modality representations. We employ a cross-self-attention mechanism to facilitate both inter- and intra-modal information transfer, while using graph convolution to incorporate updated features into item and product modal representations. Experimental results on three public datasets demonstrate MR-CSAF outperforms eight baseline methods, validating its effectiveness in providing personalized recommendations, advancing the field of personalized recommendation in complex multimodal environments.

Copy number variation (CNV) serves as a crucial contributor to genetic diversity, exerting a profound influence on phenotypic diversity, traits of economic significance, and the evolutionary trajectory of livestock species. This study aimed to dissect the genome-wide CNV landscape of the Nanyang cattle line (Nanyang, Pinnan, and Xianan cattle) to identify functionally relevant CNVs associated with key economic traits and breed differentiation. In this study, 27 resequencing datasets were utilized to analyze the genome-wide distribution of CNVs in three breeds of Nanyang cattle (Nanyang cattle, Pinnan cattle, and Xianan cattle) based on the latest reference genome ARS-UCD2.0. This study identified a total of 97,564 CNVs, and after merging CNVs with overlapping genomic positions, we obtained 10,349 CNV regions (CNVRs), accounting for 1.48% of the reference genome. Functional enrichment analysis showed that CNVR genes were mainly involved in organ development, neural regulation, immune regulation, and metabolism. In addition, 131 CNVRs overlapped with 81 quantitative trait loci (QTLs), such as growth and carcass QTL, multiple birth QTL, tenderness score QTL, and antal follicle number QTL. Additionally, AOX1, KRT72, and ZBTB7C were found to overlap with body weight QTLs. Furthermore, a selective sweep analysis of CNVR revealed that numerous genes (KIF26A, SPINT4, OR5W1, etc.) exhibited divergent copy numbers between breeds. Conclusively, this study facilitates comprehension of the genetic characteristics of the Nanyang cattle line at the CNV level and furnishes valuable information for the advancement of the Nanyang cattle line breeding system.

Gene regulatory network (GRN) inference is a central task in systems biology. However, due to the noisy nature of gene expression data and the diversity of regulatory structures, accurate GRN inference remains challenging. We hypothesize that integrating multi-source features and leveraging an attention mechanism that explicitly captures graph structure can enhance GRN inference performance. Based on this, we propose GTAT-GRN, a deep graph neural network model with a graph topological attention mechanism that fuses multi-source features. GTAT-GRN includes a feature fusion module to jointly model temporal expression patterns, baseline expression levels, and structural topological attributes, improving node representation. In addition, we introduce the Graph Topology-Aware Attention Network (GTAT), which combines graph structure information with multi-head attention to capture potential gene regulatory dependencies. We conducted comprehensive evaluations of GTAT-GRN on multiple benchmark datasets and compared it with several state-of-the-art inference methods, including GENIE3 and GreyNet. The experimental results show that GTAT-GRN consistently achieves higher inference accuracy and improved robustness across datasets. These findings indicate that integrating graph topological attention with multi-source feature fusion can effectively enhance GRN reconstruction.

Machine learning has attracted much attention in the field of genomic prediction due to its powerful predictive capabilities, yet the lack of an explanatory nature in modeling decisions remains a major challenge. In this study, we propose a novel machine learning method, ExAutoGP, which aims to improve the accuracy of genomic prediction and enhance the transparency of the model by combining automated machine learning (AutoML) with SHapley Additive exPlanations (SHAP). To evaluate ExAutoGP’s effectiveness, we designed a comparative experiment consisting of a simulated dataset and two real animal datasets. For each dataset, we applied ExAutoGP and five baseline models—Genomic Best Linear Unbiased Prediction (GBLUP), BayesB, Support Vector Regression (SVR), Kernel Ridge Regression (KRR), and Random Forest (RF). All models were trained and evaluated using five repeated five-fold cross-validation, and their performance was assessed based on both predictive accuracy and computational efficiency. The results show that ExAutoGP exhibits robust and excellent prediction performance on all datasets. In addition, the SHAP method not only effectively reveals the decision-making process of ExAutoGP and enhances its interpretability, but also identifies genetic markers closely related to the traits. This study demonstrates the strong potential of AutoML in genomic prediction, while the introduction of SHAP provides actionable biological insights. The synergy of high prediction accuracy and interpretability offers new perspectives for optimizing genomic selection strategies in livestock and poultry breeding.

Introduction: The assessment of the severity of fruit disease is crucial for the optimization of fruit production. By quantifying the percentage of leaf disease, an effective approach to determining the severity of the disease is available. However, the current prediction of disease degree by machine learning methods still faces challenges, including suboptimal accuracy and limited generalizability.Methods: In light of the growing application of large model technology across a range of fields, this study draws upon the DINOV2 visual large vision model backbone network to construct the DINOV2-Fruit Leaf Classification and Segmentation Model (DINOV2-FCS), a model designed for the classification and severity prediction of diverse fruit leaf diseases. DINOV2-FCS employs the DINOv2-B (distilled) backbone feature extraction network to enhance the extraction of features from fruit disease leaf images. In fruit leaf disease classification, for the problem that leaf spots of different diseases have great similarity, we have proposed Class-Patch Feature Fusion Module (C-PFFM), which integrates the local detailed feature information of the spots and the global feature information of the class markers. For the problem that the model ignores the fine spots in the segmentation process, we propose Explicit Feature Fusion Architecture (EFFA) and Alterable Kernel Atrous Spatial Pyramid Pooling (AKASPP), which improve the segmentation effect of the model.Results: To verify the accuracy and generalizability of the model, two sets of experiments were conducted. First, the labeled leaf disease dataset of five fruits was randomly divided. The trained model exhibited an accuracy of 99.67% in disease classification, an mIoU of 90.29%, and an accuracy of 95.68% in disease severity classification. In the generalizability experiment, four disease data sets were used for training and one for testing. The mIoU of the trained model reached 83.95%, and the accuracy of disease severity grading was 95.24%.Discussion: The results demonstrate that the model exhibits superior performance compared to other state-of-the-art models and that the model has strong generalization capabilities. This study provides a new method for leaf disease classification and leaf disease severity prediction for a variety of fruits. Code is available at https://github.com/BaiChunhui2001/DINOV2-FCS.