Introduction: Drug-drug interactions (DDIs) represent a substantial challenge in contemporary pharmacotherapy, especially given polypharmacy, the effects of foods, and the modification of host-microbiota systems on drugs. Although useful, existing DDI identification techniques have many constraints related to cost, time, and scalability. Methods: A literature review was conducted using PubMed, Scopus, Web of Science, and IEEE Xplore, focusing on machine learning techniques, deep neural architectures, and network-based models that integrate multi-omic, pharmacological, and clinical data. Results: By combining chemical, biological, and clinical data into scalable computer platforms, demonstrated that artificial intelligence techniques, such as machine learning (ML) and deep learning (DL), are changing the prediction of DDI. Some notable studies, such as DeepDDI, TP-DDI, and Decagon, use approaches that successfully capture the intricate PK-PD interactions of pharmaceuticals. On the other hand, food-drug interactions and microbiome-mediated drug interactions were also successfully predicted using multimodal and graph-based models, respectively. Discussion: Critical issues, such as insufficient data, class imbalance, and model interpretability, must be addressed through explainable AI and multimodal fusion techniques. Conclusion: The purpose of this article is to present an overview of how artificial intelligence might serve not only as a tool but also as a strategic solution for safe prescribing and tailored pharmacotherapy, hence opening up new avenues for the field of drug safety science.
Abstract: Lysosomal Storage Diseases (LSDs) are a group of rare genetic disorders characterized by impaired lysosomal degradation leading to the accumulation of undegraded macromolecules due to defective lysosomal function. Despite advances, challenges include timely diagnosis, personalized intervention, and treatment optimization. Artificial Intelligence (AI) is becoming a transformative part of precision medicine by addressing limitations via new diagnostic and pharmacological tools. This review examines AI applications in LSD management, from diagnostic accuracy through imaging and multi-omics, to drug discovery optimization. A comprehensive literature search was conducted using databases such as PubMed, Scopus, Web of Science, and Google Scholar. The search combined controlled vocabulary and free-text terms (e.g., "lysosomal storage diseases," "artificial intelligence," "machine learning," and "drug discovery"). Studies published in English over the past 10 years were included based on relevance, clinical applicability, and methodological rigor. Studies without AI-LSD applications or experimental validation were excluded. AI demonstrates potential through explainable AI, federated learning, and digital twins. Integration with wearable sensors and gene editing tools shows promise for monitoring and treatment prediction in LSD care. However, considerable barriers exist regarding data standardization, algorithmic bias, and ethical concerns. While AI has demonstrated potential in enhancing diagnostic accuracy, integrating multi-omics data, and optimizing drug discovery, challenges such as data standardization, algorithmic bias, and regulatory compliance must be addressed for broader clinical adoption. Future efforts require interdisciplinary collaboration, robust research frameworks, and regulatory policies to facilitate ethical and effective AI-driven precision medicine solutions in LSD treatment.
Background: This scoping review summarizes the types and applications of Artificial Intelligence (AI) technologies for predicting responses to Rheumatoid Arthritis (RA) treatment. Methods: A detailed search of PubMed and Google Scholar was performed from January, 2021 to August, 2025 to identify cohort studies using AI for predicting RA treatment responses. Information on the study design, AI strategies, data sources, and model performance was mined and descriptively synthesized. Results: A total of 805 articles were initially identified, of which 46 met the inclusion criteria. Logistic regression, support vector machines, random forest, and ensemble models were the most commonly used AI models. Various data sources, such as clinical parameters and genomic, transcriptomic, proteomic, and electronic health record data, have also been used. Most of the models showed moderate-to-high discriminative performance (AUROC > 0.80). Discussion: This review reveals the possibilities of AI in predicting RA treatment responses. Models indicated high accuracy, particularly when multi-omics data were used. Nevertheless, there was considerable variation in the performance of models in different studies as they used different data sources, outcome definitions, and validation methods. These drawbacks restrict the comparability and generalizability of the existing models, and real-world issues of interpretability and applicability remain. Conclusion: AI has demonstrated significant potential for predicting responses to RA treatment and assisting in individualized therapy. Further studies are needed focusing on standardized performance assessment, multi-center external validation, and multi-omics combined data to create clinically reliable and generalizable AI models.
Introduction Antimicrobial peptides (AMPs) are pivotal for developing novel antibiotics to combat escalating drug resistance, yet current computational prediction methods lack sufficient precision for reliable drug discovery.Methods The study introduces a DeepALM (Deep Antimicrobial peptide Learning Model), a context-aware deep learning framework that leverages Natural Language Processing (NLP) to enhance AMP classification. It integrates Text Convolutional Neural Networks (TextCNN), Bidirectional Long Short-Term Memory Networks (BiLSTM), and Attention Mechanisms into a novel TextCNN-BiLSTM-Attention architecture that captures contextual sequence features.Results Evaluated on benchmark datasets, DeepALM achieves an accuracy of 91.63%, outperforming established models like Transformer and BERT. It also provides interpretable insights into the biological and physicochemical properties that drive predictions.Discussion DeepALM's performance benefits from BiLSTM's bidirectional sequence dependency capture and Attention's focus on key regions. While it aligns with features like hydrophilicity and charge, exploring properties such as the isoelectric point remains a future direction. Integrating structural analysis and multimodal frameworks is proposed to advance the field further.Conclusion DeepALM represents a significant advancement in AMP classification, with high accuracy and interpretability. It serves as a robust tool for AMP-based drug development, and future work will integrate physicochemical and structural analyses to enhance utility.
Abstract: Artificial intelligence (AI) is booming and has applications in almost every domain today. With efficient use of AI techniques, biomedical problems are being solved in a comparatively shorter time frame compared to traditional clinical experiments. This paper reviews and presents a detailed study of applications of AI in Network Pharmacology (NP). This review encompasses a number of studies published recently in the field of Network Pharmacology and Artificial Intelligence. The papers were selected by searching the following keywords on the Google Scholar database: Network Pharmacology, and AI in combination with Chinese medicine, Ayurveda, Drug mechanisms, Drug Prediction, etc. Recent studies related to various methods and approaches for the use of AI in Network Pharmacology were selected. After a comprehensive review, it was observed that Artificial Intelligence, Network Pharmacology, and Machine Learning are widely used in drug development by creating networks of biological entities and inferring connections among them. AI in network pharmacology is used for targeting various problems, including drug discovery, drug-target interaction prediction, personalized medicine, drug safety, natural products and Ayurveda, traditional Chinese medicine, prediction of drug mechanisms, and many others. Artificial Intelligence techniques such as Multiple Linear Regression (MLR), Logistic Regression, Decision Trees, Support Vector Machines (SVM), and Principal Component Analysis (PCA) are widely used for early-stage data analysis and screening. This review comprehensively integrates major methods and datasets used in network pharmacology and summarizes the use of algorithms and methods used in this domain.
Introduction: Neurodegenerative disorders such as Alzheimer’s disease, Parkinson’s disease, epilepsy, and multiple sclerosis are particularly challenging due to their complex pathophysiology and the relative lack of effective treatments Methods: Traditional drug discovery tools are costly and time-consuming, which necessitates the adoption of newer approaches. Artificial intelligence (AI) has emerged as a transformative technology in neurotherapeutics, accelerating drug discovery, drug repurposing, and personalized medicine Results: AI-driven strategies leverage extensive genomic, proteomic, and clinical trial data to identify novel drug targets, rationalize molecular design, and predict drug efficacy and toxicity. Deep learning algorithms uncover intricate biological interactions and support the identification and validation of drug candidates. AI-driven natural language processing enables automated extraction of data from the literature, thereby accelerating research. AI also facilitates drug repurposing through comprehensive analysis of large drug target networks, significantly reducing development timelines. AI driven clinical trial optimization improves patient recruitment, protocol design, and real time monitoring through predictive analytics. Discussion: Challenges such as data standardization, regulatory compliance, and model interpretability must be addressed to ensure effective integration of AI into drug development. Continued advances in AI, automation, robotics, and quantum computing are expected to further refine neurological drug discovery and personalized therapeutic approaches. Autonomous laboratories integrating AI with high throughput screening are likely to transform neurodrug development and personalized therapy design. Conclusion: By accelerating treatment development, AI represents a paradigm shift in the fight against neurological diseases through the integration of precision medicine and real time approaches.
An error was made in the research paper titled "Elucidating the Functional Role of Predicted miRNAs in Post- Transcriptional Gene Regulation Along with Symbiosis in Medicago truncatula." There is a problem with the number as it does not match the main manuscript with the abstract, which was published in Current Bioinformatics, 2020, Vol. 15, No. 2 [1]. Details of the error and a correction are provided here. Original: Abstract: Background: microRNAs are small non-coding RNAs which inhibit translational and post-transcriptional processes whereas long non-coding RNAs are found to regulate both transcriptional and post-transcriptional gene expression. Medicago truncatula is a well-known model plant for studying legume biology and is also used as a forage crop. In spite of its importance in nitrogen fixation and soil fertility improvement, little information is available about Medicago noncoding RNAs that play important role in symbiosis. Objective: In this study we have tried to understand the role of Medicago ncRNAs in symbiosis and regulation of transcription factors. Methods: We have identified novel miRNAs by computational methods considering various parameters like length, MFEI, AU content, SSR signatures and tried to establish an interaction model with their targets obtained through psRNATarget server. Results: 149 novel miRNAs are predicted along with their 770 target proteins. We have also shown that 51 of these novel miRNAs are targeting 282 lncRNAs. Conclusion: In this study role of Medicago miRNAs in the regulation of various transcription factors are elucidated. Knowledge gained from this study will have a positive impact on the nitrogen fixing ability of this important model plant, which in turn will improve the soil fertility. Corrected: Abstract: Background: Non-coding RNAs (ncRNAs) are important regulators of gene expression. MicroRNAs (miRNAs) are small ncRNAs, which inhibit translational and post-transcriptional processes, whereas long ncRNAs regulate both transcriptional and post-transcriptional gene expression. Medicago truncatula is a well-known model plant for studying legume biology. Despite its importance in nitrogen fixation, little information is available about Medicago ncRNAs that play an important role in symbiosis. Objective: This study aims to predict the miRNAs and their targets from the genome of M. truncatula and elucidate their roles in symbiosis and transcriptional regulation. Methods: We have developed a computational method to identify miRNAs in M. truncatula. In addition, we have also predicted the targets of these miRNAs involved in various biological processes and established an interaction model for symbiosis. Results: We have predicted 186 miRNAs, of which 165 are novel. Additionally, 770 proteins targeted by these miRNAs are also predicted. We show that 51 of these novel miRNAs target 282 lncRNAs. The symbiosis-related gene regulation is explored through analyzing the interactions between predicted miRNAs and their nodulin target proteins. Conclusion: Knowledge gained from this study will enhance our understanding of the nitrogen-fixing ability of this important model plant, which in turn will improve soil fertility. We regret the error and apologize to readers. The original article can be found online at https://www.eurekaselect.com/article/101155
The incorporation of multi-omics strategies, namely genomics, transcriptomics, proteomics, metabolomics, and epigenomics, has been instrumental for promoting crop improvement by providing comprehensive views of the molecular processes driving complex agricultural traits, including enhanced stress tolerance, yield, and nutritional quality. This review presents an overview of the computational methods and tools currently used to analyze and integrate multi-omics data in crops. We then systematically classify them according to integrative strategies (early, intermediate, and late), and analytical methodologies (statistical, machine learning, network-based). Recent advancements in deep learning and explainable AI for predictive trait modeling are highlighted. It also discusses key knowledge gaps, including the under-representation of minor and climate-resilient crops, as well as challenges posed by data heterogeneity, scalability, and field-level validation. Through a newly proposed classification and evaluation framework, the aim of this review is to provide guidelines for researchers to choose computational pipelines and pave the way for future research on data-driven crop improvement and sustainable agriculture.
Artificial Intelligence (AI) aims to develop intelligent models that enhance the implementation of smart devices by combining science and engineering. This amalgamation has the potential to revolutionize traditional techniques, ushering in a new era of AI-driven solutions. These techniques are instrumental in leveraging knowledge, refining decision-making processes, and tackling complex problems. In healthcare, AI is progressively reshaping the landscape, introducing novel techniques, therapies, diagnostics, and economic models. Within the pharmacy industry, AI is rapidly enhancing efficiency and accuracy while minimizing quality-related issues. Its applications includes drug discovery, formulation, manufacturing, product development, diagnosis, clinical trials, quality assurance and control, as well as research and development. AI algorithms play a pivotal role in analyzing vast medical datasets, pinpointing drug targets, and optimizing drug design. Key AI techniques include machine learning, deep learning, Artificial Neural Networks (ANNs), Deep Neural Networks (DNNs), and Recurrent Neural Networks (RNNs), with associated tools widely employed in pharmacy settings. Moreover, in disease diagnosis and therapy, AI facilitates informed decision- making by leveraging patient medication histories and current disease conditions to tailor accurate and personalized treatment plans. This article comprehensively covers the myriad applications of AI in the pharmacy field. Overall, AI holds the potential to revolutionize pharmacy practice, enhancing product quality, efficacy, and patient outcomes through precise, effective, and personalized medication management.
Understanding the genetic basis of cancer requires the accurate identification of driver genes and driver mutations, those alterations that promote tumorigenesis, while distinguishing them from neutral, or passenger, mutations. This review provides a comprehensive overview of computational strategies developed to detect and prioritise cancer drivers at both the gene and mutation levels. The review systematically classifies and compares more than 20 widely used tools, highlighting differences in their conceptual foundations, including sequence-based, structure-based, statistical, machine learning, and network/pathway-based methods. These tools leverage diverse types of data, including mutation frequency and evolutionary conservation, as well as gene expression profiles and interaction networks, to assess the functional relevance of somatic alterations. By integrating complementary approaches, researchers can enhance the sensitivity and specificity of driver prediction, particularly in cases involving rare or heterogeneous mutations. This review aims to serve as a practical guide for researchers and clinicians seeking to apply or evaluate current methods for cancer driver identification.
Introduction The rapid spread of false pandemic-related news on social media poses a serious threat to public health. Existing detection methods face challenges, such as unreliable edge connections and static network assumptions. This study aimed to develop a robust and adaptive framework for detecting fake pandemic news by addressing these limitations through dynamic interaction modeling and local feature analysis.Methods We proposed a dynamic multi-scale hypergraph neural network framework. It employs a hypergraph neural network to capture complex interactions among user groups across time and a cross-time fusion mechanism to integrate temporal dynamics. Local user behavior features are extracted using propagation tree forests. Results: Experiments on multiple datasets showed that the proposed framework significantly outperformed existing propagation-based methods, achieving higher accuracy and F1 scores.Methods We proposed a dynamic multi-scale hypergraph neural network framework. It employs a hypergraph neural network to capture complex interactions among user groups across time and a cross-time fusion mechanism to integrate temporal dynamics. Local user behavior features are extracted using propagation tree forests. Results: Experiments on multiple datasets showed that the proposed framework significantly outperformed existing propagation-based methods, achieving higher accuracy and F1 scores.Discussion Although MSD-HNN achieved promising results, its reliance on textual features and the limited scale of certain datasets may affect generalizability. Future work will focus on incorporating multimodal data, enhancing model interpretability, and improving computational efficiency for real-time deployment.Conclusion The dynamic multi-scale hypergraph neural network framework effectively detects fake pandemic news by leveraging global and local features. Future work will explore incorporating multimodal data to improve model generalization and adaptability.
Background The rapid advancement of precision medicine in oncology has intensified the need for the accurate prediction of cancer drug responses. Deep learning technologies present a promising pathway toward addressing this challenge. However, because cancer is a complex disease influenced by multiple molecular layers, the integration of multi-omics data and the elucidation of their intricate relationships are essential for improving predictive accuracy. Traditional methods currently have limitations in fully exploiting multi-omics information and capturing complex interactions, highlighting the need for more sophisticated approaches.Objective This study aims to design a new method for integrating drug and cell line data through cross-attention, thereby improving the accuracy of drug response prediction.Methods In this study, we leverage a Transformer to encode drug SMILES sequences to capture fine-grained molecular semantics and introduce a cross-attention mechanism to model bidirectional interactions between drugs and multi-omics features, thereby significantly improving the accuracy and robustness of drug response prediction.Results Experimental evaluations demonstrate that, compared with traditional approaches, the proposed method achieves considerable improvements in accuracy and stability across diverse prediction scenarios. The model exhibits robust performance in managing complex multi-omics inputs and reliably predicting drug responses.Discussion The proposed AttenDRP model demonstrates strong potential in drug response prediction. Future research can expand its applicability by leveraging large and diverse datasets and incorporating patient-derived or clinical data to further strengthen robustness and translational utility.Conclusion AttenDRP integrates heterogeneous drug features with multi-omics cell line data through a cross-attention network, enabling the modeling of fine-grained interactions. This approach improves the accuracy and robustness of drug response prediction and provides a methodological foundation for advancing precision oncology.
Introduction Researchers have found that initialization methods can accelerate convergence speed and reduce training loss in graph convolutional neural networks for identifying anti-cancer peptides (ACPs). However, existing initialization methods for odd activation functions lack a reasonable mathematical explanation. Meanwhile, some initialization methods require the activation function's derivative to be equal to 1 near the origin.Methods In this study, we proposed a novel initialization method (NPI) with theoretical derivation to explain the stability of an odd activation function. Based on the defined non-odd activation function KPReLU, we proved through mathematical derivation that NPI maintains the stability of the mean and variance of weight matrices during forward and backward propagation in graph neural networks, even when the derivative of the activation function near the origin deviates from 1. By integrating NPI, we developed a model named GNN-NPI, which uses multi-layered structures to extract and integrate information at different levels while considering both sequential and graph features.Results Experimental results on the benchmark dataset show that GNN-NPI outperforms other algorithms with SN 94.87%, SP 92.65%, AVE 93.76%, ACC 93.83%, and MCC 0.88. Moreover, GNN-NPI significantly surpasses state-of-the-art models, with improvements of 11.49%, 2.56%, 2.05%, and 0.06 in SN, AVE, ACC, and MCC, respectively, on the independent dataset.Discussion GNN-NPI enhances training efficiency and generalization. Feature selection and architectural optimization ensure robust discriminative capability.Conclusion The superior performance indicates that GNN-NPI is effective for ACP recognition. The powerful capability of NPI may contribute to research in biology and bioinformatics. The dataset and codes are available at .
Introduction MicroRNAs (miRNAs) are crucial in regulating gene expression. Identifying miRNA-disease association (MDA) is essential for disease diagnosis and treatment. Due to the high cost of traditional biological experiments, computational models have emerged as an effective alternative approach for MDA prediction. However, existing methods often lack flexibility in feature fusion and processing and cannot capture long-range dependencies, thereby limiting the accuracy of predictions.Methods In this paper, we propose a novel MDA prediction model, CGDAMDA, which integrates an adaptive content-guided fusion mechanism with a DilateFormer architecture. First, homologous information of miRNAs and diseases is independently fused via the adaptive content-guided fusion mechanism to generate comprehensive similarity representations. A miRNA-disease heterogeneous network is then constructed, followed by Laplacian positional encoding and Weisfeiler-Lehman absolute role encoding. Deep feature representations are extracted using a combination of sliding-window attention and multiscale dilated attention, enabling the model to capture complex feature dependencies. Finally, the learned miRNA and disease embeddings are fed into an XGBoost classifier to predict potential MDAs.Results In 5-fold cross-validation, CGDAMDA achieves AUC scores of 0.9585 +/- 0.0041 and 0.9632 +/- 0.0021 on the HMDD v2.0 and HMDD v3.2 datasets, respectively, outperforming several state-of-the-art models.Discussion CGDAMDA improves MDA prediction by introducing adaptive fusion and DilateFormer encoder. In the future, consideration is given to integrating multi-omics data to further enhance the generalization ability of the model.Conclusion In conclusion, our proposed CGDAMDA obtains high-quality node feature representations through adaptive feature fusion and shallow-to-depth feature extraction, which effectively improves the prediction performance of MDA.
Introduction: The analysis of the human microbiome has gained relevance in biomedical research due to its association with various diseases, including cancer. Understanding the human microbiome’s role in health and disease requires robust analytical strategies capable of addressing the complexity and variability of taxonomic abundance data. This study presents a novel machine learning framework for classifying human microbiome profiles, integrating alpha diversity metrics based on Hill numbers, a dissimilarity-driven selection of representative subsets, and a heterogeneous ensemble learning architecture based on stacking. Methods: The method enhances generalization performance by leveraging the probabilistic outputs of multiple base classifiers—Random Forest, K-Nearest Neighbors, Support Vector Machine, Gradient Boosting, and Multi-Layer Perceptron—combined via a logistic regression meta-model. The workflow incorporates internal stratified cross-validation to prevent data leakage and applies a rigorous experimental design comprising 25 independent iterations. Results: The proposed approach outperforms traditional classification baselines, achieving an average sensitivity of 60% and a balanced precision-recall performance, underscoring its utility in clinical settings where early detection is critical. Discussion: This study underscores how methodological choices in diversity representation and ensemble design can critically influence predictive performance and reproducibility. Conclusion: This work demonstrates that incorporating alpha diversity metrics and ensemble methods provides a powerful tool for advancing microbiome-based diagnostics and supports the integration of machine learning into personalized medicine initiatives.
Introduction: Rice blast, caused by the fungal pathogen Magnaporthe oryzae, poses a significant threat to global rice production. Conventional methods for identifying resistance-associated genes are often laborious and time-consuming. Machine learning has proven effective in gene prioritization for complex traits, especially in human genomics, yet its application in plant immunity remains limited. To address this gap, we present CGAN-SAE, a novel deep learning framework that integrates a Conditional Generative Adversarial Network (CGAN) with a Sparse Autoencoder (SAE) to predict candidate genes involved in rice immune responses. The approach enables data-efficient and interpretable modeling, offering a powerful strategy to accelerate the development of blast-resistant rice varieties. Methods: The CGAN-SAE framework combines a Conditional Generative Adversarial Network (CGAN) to generate biologically plausible synthetic gene expression profiles, thereby alleviating limitations imposed by small sample sizes, with a Gated Axial-Attention-enhanced Stacked Autoencoder (SAE) for robust feature extraction and global pattern recognition. Principal Component Analysis (PCA) is then applied to interpret the learned latent features and evaluate gene-level importance. Results: Transcriptomic analysis identified 891 differentially expressed genes (DEGs). Using CGAN-SAE, we prioritized five high-confidence candidate genes potentially associated with disease resistance. The model achieved an AUC of 0.80, accuracy of 0.75, F1-score of 0.77, precision of 0.83, and recall of 0.71, outperforming multiple baseline and state-of-the-art models. These results demonstrate its superior ability to extract meaningful biological signals from limited transcriptomic data and identify key regulatory factors. Discussion: CGAN-SAE represents a significant advancement in deep learning-based gene prioritization under data-scarce conditions. Although performance depends on the quality of input DEGs and may require species-specific tuning, the framework shows strong potential for extension to other crops such as maize and soybean, and even to human disease genomics. Future integration of protein-protein interaction networks and multi-omics data could further improve its biological interpretability and broad applicability across host-pathogen systems. Conclusions: This study demonstrates that deep learning-driven gene prioritization is feasible and effective in plant disease resistance research, offering a scalable framework for candidate gene discovery under limited data conditions.
Introduction: Olfaction is a complex process governed by genes that are tightly regulated by Transcription Factors (TFs). Although a large number of human Olfactory Receptor (OR) genes have been identified, the regulatory mechanisms controlling their expression and function remain largely unknown. Understanding these regulatory networks is critical for studying human health, sensory perception, and disease mechanisms. Methods: We developed the Olfactory Receptor Transcription Regulation Database (ORTRD), a manually curated resource compiling 7,139 causal TF-olfactory gene interactions. Data were integrated from multiple sources, including TF binding information, gene expression profiles, and functional annotations. Each interaction was categorized according to its regulatory effect (activation or repression), using experimental evidence and computational inference. The database is stored in MySQL and designed for periodic updates as new validated data becomes available. Results: ORTRD contains detailed information on human OR genes, associated TFs, regulatory signs, log2 Fold Changes (log2FC), and TF binding motifs. Interactions can be visualized as signed directed graphs that represent activation or repression relationships within Gene Regulatory Networks (GRNs). Users can query the database to retrieve and download specific data subsets for customized analyses. Discussion: By combining experimentally validated interactions with functional annotations, ORTRD provides a high-coverage resource for interpreting GRNs in human olfaction. It allows researchers to explore transcriptional regulation with confidence and supports studies in genomics, physiology, and sensory biology. Conclusion: ORTRD is a comprehensive, user-friendly resource for studying transcriptional regulation of human olfactory receptors. It is publicly accessible at and will be updated periodically to incorporate new validated data.
Introduction Type 2 diabetes mellitus (T2DM) and cardiovascular diseases (CVDs) are major non-communicable disorders that significantly contribute to global morbidity and mortality, particularly in the North Indian population. This study aims to leverage machine learning (ML) to assess the risk of developing T2DM and CVDs using demographic, biochemical, and lifestyle parameters collected from an Indian cohort.Methods A cross-sectional dataset comprising clinical and biochemical features was analyzed using supervised ML algorithms, including Support Vector Machine (SVM), Random Forest (RF), Logistic Regression (LR), and Extreme Gradient Boosting (XGBoost). Feature normalization, correlation analysis, and hyperparameter tuning were performed to optimize model performance. Models were evaluated using accuracy, precision, recall, F1-score, and area under the receiver operating characteristic curve (AUC).Results The ML models effectively classified individuals at high or low risk of T2DM and CVD. Ensemble-based models, such as Random Forest and XGBoost, achieved superior predictive performance compared to baseline algorithms, indicating their suitability for population-level screening and early risk identification.Discussion The findings highlight the potential of ML in developing low-cost, data-driven decision-support tools for early identification of chronic disease risk. Population-specific modeling and feature interpretability are essential for improving generalizability and clinical translation of predictive systems.Conclusion This study establishes an interpretable ML-based framework capable of predicting T2DM and CVD risk among North Indian individuals. The approach may support precision-prevention strategies and guide targeted public-health interventions.
Introduction: Cancer is the second leading cause of death worldwide. Although substantial efforts have been devoted to developing effective treatments, conventional chemotherapy and radiotherapy remain limited by systemic toxicity, drug resistance, and high recurrence rates. Anticancer agents offer an alternative therapeutic approach, yet experimental discovery strategies are expensive and time-consuming. Methods: This study developed a computational model, AntiCanNet, to predict Anti-Cancer Small Molecules (ACSMs). Molecular features were generated using PaDEL and ChemGPT, a chemical large language model. A small-molecule network was constructed based on structural similarity. These features and the network were processed through a graph neural network to obtain high-level representations, followed by a fully connected neural network for prediction. Results: AntiCanNet was evaluated on one training dataset and two test datasets. Cross-validation on the training dataset yielded an AUC of 0.971, and both test datasets achieved AUC values above 0.9, demonstrating strong predictive performance Discussion: Ablation analyses supported the effectiveness of the model design. Several latent ACSMs identified by AntiCanNet showed potential associations with cancer-related pathways. Conclusion: AntiCanNet provides an efficient computational approach for identifying ACSMs and may facilitate the discovery of previously unrecognized anti-cancer agents.
Introduction Children with Coronavirus Disease 2019 (COVID-19) and those with Multisystem Inflammatory Syndrome in Children (MIS-C) exhibit similar inflammatory responses, yet distinct differences exist, particularly in immune cell reactions. This study aimed to uncover the differences between them.Methods This study analyzed plasma cell-free ribonucleic acid [cfRNA] and whole blood RNA [wbRNA] from children with COVID-19, MIS-C, and a healthy control group. Pediatric blood and plasma samples were collected from three hospital systems, including patients with PCR-confirmed COVID-19 and those meeting CDC-defined criteria for MIS-C. COVID-19 cases required SARS-CoV-2 positivity within 14 days of sampling, while MIS-C diagnoses were adjudicated by multidisciplinary teams based on clinical and inflammatory features. Each sample was represented by 60,708 gene expressions. Ten advanced feature ranking algorithms were first applied to yield feature lists. Then, these lists were analyzed by incremental feature selection method, which contained four classification algorithms and synthetic minority oversampling technique, to extract essential genes and build efficient prediction models and classification rules.Results Several important genes were discovered, which were identified by multiple feature ranking algorithms. The optimal models on cfRNA and wbRNA achieved weighted F1 scores exceeding 0.9.Discussion Analysis of the cfRNA dataset revealed low EPSTI1 expression and low SNHG6 expression in MIS-C patients, while the wbRNA dataset indicated high IFI27 expression in COVID-19 patients and high JUN expression in COVID-19 and MIS-C patients compared with non-inflammatory controls.Conclusion The newly found genes can serve as potential qualitative markers for COVID-19 or MIS-C, expanding the scope of biomarkers for COVID-19 or MIS-C. These findings will be helpful in investigating the pathogenic mechanisms of MIS-C.
