Health emergencies need to face the potential challenges about the public psychological healthcare and heart disease concurrent occurrence probability, especially for the large-scale pandemic disease spread situation. The complex bidirectional relationship between cardiovascular and psychological diseases highlights the importance of appropriate combination drug strategies, because of psychological therapy medicines could affect the heart sympathetic and parasympathetic nerve responses, at the same time, there may also cause a counter-effect. However, existing drug interaction prediction models focus only on the two-by-two drug relationship, which is difficult to meet the actual clinical needs for synergistic combination drugs. To address this, this paper proposes a Loewe score dual-guided diffusion model for synergistic drug combination generation aimed at generating drug combinations with high synergistic potential. It introduces drug text description information and Loewe synergistic score information into the training process of diffusion models, uses Loewe scores for conditional bias and weighted loss, and doubleguided denoising network learning to generate drug combinations with high synergy potential. The experimental results show that our proposed model is able to generate reasonable and reliable drug combinations, which provides a new perspective and an effective tool for solving the problem of clinical combination drug selection.

Identifying Systemic Lupus Erythematosus (SLE) and Antiphospholipid Syndrome (APS) has been challenging, because of the complex symptom overlap and biological differences between these two autoimmune diseases. To address this issue, this study proposes a text-based gene unsupervised clustering model and a gene dense subgraph mining algorithm to delve into their differences and similarities at the genetic level. The proposed gene unsupervised text clustering model is used for discovering the gene cross characteristics and significant differences between SLE and APS related gene groups, which is calculated by extracting semantic features from medical textual datasets. Then, the proposed gene dense subgraph mining algorithm is well identified the key and tightly connected gene sets that may play a critical role in disease mechanisms over largescale complex network, which would help to deep understand the correlations between them. The experiment results show that 10 typical functional gene groups (SLE 6 and APS 4 clusters) are clearly classified, and gave a detailed comparison analysis with the baseline methods and gave a semantic explanation for mined key related genes.

Type 2 diabetes mellitus (T2DM) requires multitargeted treatment, with drug efficacy tied to structure and function, though “structure-function divergence” exists. Drug similarity analysis and drug-target interaction (DTI) network construction are key for exploring functional associations and predicting targets. Thus, this study proposes an integrated framework for T2DM drugs that combines structural features and functional annotations, integrating similarity analysis and DTI network construction to explore links between structural features and binding affinity, aiding drug repurposing and target prediction. Via DTI networks, we calculated weighted similarity for small molecules using molecular fingerprints and functional annotations, applied global alignment with chain length weighting for double-stranded proteins, and used optimized local alignment for fusion proteins. The results indicate that the similarity between small molecule drugs of the same class is higher than that between drugs of different classes. The conservatism of core functional domains in protein drugs determines their selectivity for targets.

This study proposes a drug recommendation model based on a multi-objective optimization algorithm, aiming to provide personalized medication strategies for Systemic Lupus Erythematosus and Antiphospholipid Syndrome patients, taking into account individual differences and medication risks. Using a multi-objective optimization algorithm, the model comprehensively considers drug efficacy, safety and individual differences to achieve multi-dimensional screening of drugs. Drugs are categorized as therapeutic drugs for Systemic Lupus Erythematosus, Antiphospholipid Syndrome and other related complications, ensuring that recommendations meet different patient needs. Experimental results show that the Drug-ESIC-PC recommendation model we proposed has an accuracy of$\text{9 2 \%, } \text{8 8 \%}$, and$\text{8 4 \%}$for recommending two, three, and four drugs, respectively.

Approximate Nearest Neighbor Search (ANNS) is widely used to solve high-dimensional vector problems and has gradually become an effective solution in the field of natural language processing in recent years, including applications such as information retrieval, recommendation systems, and text similarity computation. This paper proposes an ANNSRM-SSC model to support approximate nearest neighbor search and vector retrieval over constructed large-scale multi-theme complex graphs. The model optimizes the keyword network structure by constructing a semantic keyword network and employing community detection and influential node selection to build a hierarchical index model. By integrating embedded vector calculations of semantic relevance, it quickly maps from the top-level nodes to the communities where the query nodes reside, reducing the computation of irrelevant nodes and achieving high-precision searches at low cost within the hierarchical index. Experimental analysis shows that this method achieves a best recall rate of 93.25 % in nearest neighbor search tasks, demonstrating excellent performance and providing a new solution for text retrieval.

The thickness of the Retinal Nerve Fiber Layer (RNFL) is an important indicator for glaucoma diagnosis, but it cannot be reliably assessed by conventional thickness measurement methods because glaucoma at different stages presents with different patterns of RNFL damage (e.g., early-stage glaucoma often presents with only minimal damage). For this reason, a method for glaucoma grading has been proposed based on point cloud reconstruction of the layer. The method is based on labeling Optical Coherence Tomography (OCT) slices, performing point cloud reconstruction, and extracting structural characteristics, such as point cloud count, the mean density, and the standard deviation density, from the 3-dimensional OCT point cloud data, These features are combined with a rule-driven classification strategy that uses empirical thresholds to classify the samples into three categories of Non-glaucoma, Early glaucoma, and Midadvanced glaucoma. The experimental results showed that the method demonstrated high accuracy in glaucoma grading, and in addition, the volumetric maps reconstructed from the point cloud were used as an auxiliary diagnostic tool to help clinicians assess the structural changes caused by glaucoma. The method is highly interpretable and provides important support for clinical glaucoma diagnosis.

Large-scale integration covering genetic traits, compound properties, and pharmacological mechanisms opens new avenues for drug repurposing in retinal diseases. This study presents SNF-LapDTI, a framework combining similarity network fusion (SNF) with Laplacian regularized least squares (LapRLS). SNF creates a unified similarity representation from heterogeneous sources, while LapRLS builds predictive functions in drug and target spaces to infer interaction likelihoods. Evaluations on multiple datasets confirm superiority, reaching AUROC 95.91% and AUPR 95.49% on RD.

With the rapid advancement of Brain-Computer Interface (BCI) technology, real-time control of external devices via electroencephalogram (EEG) signals has become feasible, offering new avenues for building more natural and efficient human-computer interaction systems. To address these issues, this paper proposes a Rule-based Motor Imagery Interaction Model (Rule-MII Model) that combines rule matching with multifeature fusion. By leveraging discrete wavelet transforms, timefrequency domain analysis, and symbolic pattern mining, the model extracts multidimensional EEG features and constructs a standard rule set for recognizing users' motor intentions and emotional states. The proposed model achieves classification accuracies of 91.0 % and 92.7 % on the public BCI2a and DEAP datasets for motor imagery and emotion recognition tasks, respectively. Experimental results demonstrate that the proposed method exhibits strong accuracy and responsiveness in multimodal EEG signal modeling and mixed-reality-based human-robot interaction control, providing a promising solution for immersive and context-aware human-robot collaboration.

In this paper, we proposed a protein complex functional prediction model based on biocompatibility computation, which employs an encoder-decoder architecture combined with a multi-head attention mechanism to generate a composite embedding representation that fuses multi-dimensional features. A fully-connected classifier outputs potential functional labels for the complex, thereby completing protein complex function prediction. The reliability of the model is demonstrated by the calculation of performance evaluation indicators such as m-AUPR and M-AUPR, which are expected to help in the design of safe and effective biomaterials and medical implants.

With the rapid development of Natural Language Processing technology and the growing demand for efficient information retrieval in the computer field, query optimization tasks have become an important challenge. This paper proposes a new Hierarchical Query Classification and Retrieval (HQCR) framework to improve the efficiency of query processing systems. Our proposed method achieves precise classification and processing of queries through its unique query classification mechanism and efficient hierarchical indexing strategy. The core strength of this framework lies in its ability to effectively integrate multiple features, thereby delivering outstanding performance in complex query scenarios. Experimental results show that it outperforms traditional methods in average query time and classification accuracy, verifying its effectiveness in query optimization in the computer field.

With the development of bioinformatics and machine learning techniques, protein-DNA binding site prediction has become a key topic in understanding gene regulation and molecular recognition. However, methods relying solely on sequence information often ignore important spatial distribution features in protein structures, leading to inadequate identification of complex binding sites. This paper introduces a novel protein-DNA binding site prediction method, FRD-Bind, which utilizes the Random Forest and the DRD dense clustering algorithms. By combining sequence and structural features, FRD- Bind improves the prediction accuracy. The Random Forest classifier is used to determine the predicted scores of residues, and the DRD dense clustering algorithm is then applied to reduce noise and enhance the accuracy of binding site prediction. Experimental results show that FRD-Bind achieves an accuracy rate(ACC) of 94%, which is superior to other sequence- and structure-based prediction methods. This method is particularly effective for tasks involving unbalanced positive and negative samples and helps to accurately predict protein-DNA binding sites.

With the advancement of computer vision, methods for scene image representation have greatly improved classification performance. Nevertheless, challenges remain due to the complexity of scene images, including high variability within classes and strong similarity between classes, which continue to limit performance. This study proposes a model called Mutli-Features Fusion Scene Classification(MFF-SC) model that integrates multiple dimensions of information for scene classification. By combining human skeletal and facial keypoint information, and utilizing these features with other feature information (emotions, global distribution, population density, population size, population flow characteristics) for scene classification. The fusion of features from different sources generates a unified feature vector, which is ultimately input into an SVM classifier for classification tasks. This model integrates multi-dimensional information such as images, expressions, behaviors, and environments, effectively improving the robustness and accuracy of scene classification.

The robot remote control architecture aims to reduce the computing load of the robot and improve the robot's task response speed. To address the robot's limited computational resources, this paper proposes a remote control architecture that migrates complex computations to a remote Linux server. Tasks such as face recognition, person tracking, Q&A, and product information recognition are processed remotely, and the robot executes commands received from the Linux server, thereby reducing its local computational load. At the same time, the Feature Coefficients Selection-based Decision Tree (FCS-DT) Classification Method is proposed for deployment on the remote Linux server. This approach enables real-time monitoring and optimization of system performance bottlenecks, further enhancing the efficiency of robot task processing, and in the later experiments, FCS- DT shows high accuracy and recognition speed, and the architecture is highly efficient for the computation of complex tasks of robots. Therefore, this architecture has practical value for improving the processing efficiency of robot complex tasks.

With the acceleration of global energy transition and the popularization of Electric Vehicles (EVs), regional electricity demand and EV charging price forecasting has become an important issue in power system management. In this paper, a Cross-Scale Adaptive Weighted Fusion Network(CSAWFNet) based on triple weighted fusion LSTM model is proposed to address the limitations of the existing forecasting methods in dealing with the complex and changing power demand and price fluctuations. The feature extraction was first performed using machine learning algorithms, and then as inputs to the model. The model can effectively capture the short-term high-frequency changes and long-term trends of electricity demand and price fluctuations through the improvements of sliding window, multi-scale memory fusion mechanism, adaptive memory pruning mechanism, and cross-layer connections. The experimental re-sults show that CSAWFNet outperforms traditional models in forecasting regional electricity demand and EV charging prices, particularly during peak and trough load periods and price fluctuations.

Drug-Drug Interactions (DDIs) are an issue that cannot be ignored in clinical treatment. With the improvement of people's living standards and the evolution of the disease spectrum, the relationship between emotions and the immune system has gradually attracted attention. Among them, mood swings can significantly affect the function of the immune system, and disorders of the immune system may also cause mood disorders. Therefore, the interaction between antidepressants and immunosuppressive drugs has important research value. This article deeply analyzes the interaction strength of these two types of drugs to assess their potential risks, thereby optimizing treatment plans and reducing adverse reactions. At the same time, 3D imaging technology is used to more intuitively display the binding position and action intensity of drugs at the target, providing strong support for clinical decision-making and helping to achieve safer and more personalized combined treatment plans.

Cardiovascular disease and depression often require combined use of cardiovascular and psychotropic drugs. This paper introduces Multi-Feature Fusion Learning for Discovering Synergistic Relationships (MFFL-DSR), a method to predict drug interactions. It first constructs matrices of drug features, including classification, targets, enzymes, pathways, and molecular structure. Drugs from different feature domains are then projected into a shared interaction domain. A regularization term is formulated to represent drug pairing relationships in the inter-action space, forming the MFFL-DSR objective function. Finally, iterative optimization identifies all potential drug combinations. The results show that MFFL-DSR outperforms baseline methods in six metrics: AUPR, AUC, Precision, Accuracy, Recall, and F1 score.

In this paper, we proposed a Weighted Subgraph Clustering for Protein Complex Identification (WSC-PCI) method , which combined with Gene Ontology (GO) annotations and improved Edge Clustering Coefficient (ECC) to indentify the protein complex over weighted Protein- Protein Interaction (PPI) network. By using the comprehensive weighted method to calculate the node weights within the PPI network, in order to mine the valuable and core subgraphs, which are related to the biologically significant and real protein complexes based on the core-attachment structure requirements. And then, calculating Biocompatibility value (BA-value) to identify potential proteins, which have more stable and good biocompatibility. By comparing with multiple algorithms on real datasets of different species, the results show that WSC-PCI outperforms other methods with regard to recall, precision and MMR, and the authenticity and biological significance of the identification results are verified by p-value analysis.

Healthcare data, collected from multiple sources and representing various perspectives, has become an increasingly challenging and important research topic. It permeates many aspects of evidence-based clinical practice, the healing process, and health policy formulation. However, the characteristics of redundancy, diversity, volume, inconsistency, and incompleteness make it difficult to capture valuable semantic properties and linguistic relationships. In this paper, we propose a CNN-based Bidirectional Extension of Phrase Boundary (BEPB) approach to reduce over-reliance on term frequency and mine multiple latent topical key phrases, aiming to improve the quality of key phrases in downstream Natural Language Processing (NLP) applications. The experiments indicate that the BEPB trained from unstructured user-contributed content on health social media sites has successfully mined more relevant topical phrases, particularly in the areas of clinical symptoms and co-occurrence patterns. To summarize, these findings serve as a key contributor to the advancements in Evidence-Based Medicine (EBM), paving the way for improvements in the prevention, diagnosis, treatment, and nursing of diseases.

Power load forecasting is influenced by various factors, including meteorological, economic, and social factors. Considering all influencing factors will significantly increase the model complexity, affect accuracy and prediction timeliness. In addition, as time, region, and season change, the different influence factors will also change, which could lead a great impact on accuracy to existing prediction models. In order to improve the accuracy of real-time predictive analysis, this paper proposes a Dynamic Multi Indicator Fusion (DMIF) model for processing the calculation of real-time load impact indicators and adaptively predicting and adjusting the weights of the most influential indicators to reduce complex calculations and improve real-time prediction accuracy. In the final experiment, our model showed high prediction accuracy and fast calculation speed, while reducing information redundancy in multiple indicators. Therefore, in the context of smart grids, this method has practical application value for the stable operation of power grid systems.

Question and answer system aims to empower computers to comprehend natural language and provide precise responses to user queries. To address challenges in acquiring knowledge within the realm of artificial intelligence, this paper conducts in-depth semantic association analysis on AI-related knowledge data. Based on the constructed AI knowledge graph, the paper introduces the Inter-word Association Rule Mining Query Expansion (IARM-QE) model. The model includes question classification, identification of syntactic dependencies, and answer generation. The core algorithm, CSSC, combines synonym similarity clustering and semantic similarity clustering to uncover deep semantic relationships in the data, improving the accuracy and average response time of the information retrieval system for queries.