‘Applications of machine learning in liposomal formulation and development’

Pharmaceutical Development and Technology, Journal Year: 2025, Volume and Issue: 30(1), P. 126 - 136

Published: Jan. 2, 2025

Machine learning (ML) has emerged as a transformative tool in drug delivery, particularly the design and optimization of liposomal formulations. This review focuses on intersection ML technology, highlighting how advanced algorithms are accelerating formulation processes, predicting key parameters, enabling personalized therapies. ML-driven approaches restructuring development by optimizing liposome size, stability, encapsulation efficiency while refining release profiles. Additionally, integration enhances therapeutic outcomes precision-targeted delivery minimizing side effects. presents current breakthroughs, challenges, future opportunities applying to systems, aiming improve efficacy patient various disease treatments.

Language: Английский

---

Artificial Intelligence, Computational Tools and Robotics for Drug Discovery, Development, and Delivery

Intelligent Pharmacy, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 1, 2025

Language: Английский

---

Artificial Intelligence and Internet of Things Integration in Pharmaceutical Manufacturing: A Smart Synergy

Pharmaceutics, Journal Year: 2025, Volume and Issue: 17(3), P. 290 - 290

Published: Feb. 22, 2025

Background: The integration of artificial intelligence (AI) with the internet things (IoTs) represents a significant advancement in pharmaceutical manufacturing and effectively bridges gap between digital physical worlds. With AI algorithms integrated into IoTs sensors, there is an improvement production process quality control for better overall efficiency. This facilitates enabling machine learning deep real-time analysis, predictive maintenance, automation—continuously monitoring key parameters. Objective: paper reviews current applications potential impacts integrating concert technologies like cloud computing data analytics, within sector. Results: Applications discussed herein focus on industrial analytics quality, underpinned by case studies showing improvements product reductions downtime. Yet, many challenges remain, including ethical implications AI-driven decisions, most all, regulatory compliance. review also discusses recent trends, such as drug discovery blockchain traceability, intent to outline future autonomous manufacturing. Conclusions: In end, this points basic frameworks that illustrate ways overcome existing barriers increased efficiency, personalization, sustainability.

Language: Английский

---

Efficient substructure feature encoding based on graph neural network blocks for drug-target interaction prediction

Frontiers in Pharmacology, Journal Year: 2025, Volume and Issue: 16

Published: March 5, 2025

Predicting drug-target interaction (DTI) is a crucial phase in drug discovery. The core of DTI prediction lies appropriate representations learning and target. Previous studies have confirmed the effectiveness graph neural networks (GNNs) compound feature encoding. However, these GNN-based methods do not effectively balance local substructural features with overall structural properties molecular graph. In this study, we proposed novel model named GNNBlockDTI to address current challenges. We combined multiple layers GNN as GNNBlock unit capture hidden patterns from within ranges. Based on GNNBlock, introduced enhancement strategy re-encode obtained features, utilized gating units for redundant information filtering. To simulate essence that only protein fragments binding pocket interact drugs, provided encoding target using variant convolutional networks. Experimental results three benchmark datasets demonstrated highly competitive compared state-of-the-art models. Moreover, case study candidates ranking against different targets affirms practical GNNBlockDTI. source code available at https://github.com/Ptexys/GNNBlockDTI.

Language: Английский

---

Hallmarks of artificial intelligence contributions to precision oncology

Nature Cancer, Journal Year: 2025, Volume and Issue: unknown

Published: March 7, 2025

Language: Английский

---

Structural Virology: The Key Determinants in Development of Antiviral Therapeutics

Viruses, Journal Year: 2025, Volume and Issue: 17(3), P. 417 - 417

Published: March 14, 2025

Structural virology has emerged as the foundation for development of effective antiviral therapeutics. It is pivotal in providing crucial insights into three-dimensional frame viruses and viral proteins at atomic-level or near-atomic-level resolution. Structure-based assessment components, including capsids, envelope proteins, replication machinery, host interaction interfaces, instrumental unraveling multiplex mechanisms infection, replication, pathogenesis. The structural elucidation enzymes, proteases, polymerases, integrases, been essential combating like HIV-1 HIV-2, SARS-CoV-2, influenza. Techniques X-ray crystallography, Nuclear Magnetic Resonance spectroscopy, Cryo-electron Microscopy, Tomography have revolutionized field significantly aided discovery ubiquity chronic infections, along with emergence reemergence new threats necessitate novel strategies agents, while extensive diversity their high mutation rates further underscore critical need analysis to aid development. This review highlights significance structure-based investigations bridging gap between structure function, thus facilitating therapeutics, vaccines, antibodies tackling emerging threats.

Language: Английский

---

Enhancing Drug Discovery via Physics-Guided Deep Generative Models

Communications in computer and information science, Journal Year: 2025, Volume and Issue: unknown, P. 16 - 30

Published: Jan. 1, 2025

Language: Английский

---

Kinase Drug Discovery: An Artificial Intelligence Revolution

AI in Precision Oncology, Journal Year: 2025, Volume and Issue: unknown

Published: March 28, 2025

Language: Английский

---

Assessing the Robustness and Scalability of Machine Learning Methods to Accelerate Ultralarge High-Throughput Docking Campaigns

ACS Omega, Journal Year: 2025, Volume and Issue: unknown

Published: April 6, 2025

Structure-based virtual screening methods are, nowadays, one of the key pillars computational drug discovery. In recent years, high-throughput docking campaigns aided by machine learning (ML)-based protocols have emerged as a way to accelerate identification top-scoring molecules within ultralarge chemical molecule libraries. However, studies validating these ML approaches used or two targets and/or small Herein, we extended validation at retrieving hits in an accelerated fashion using standard publicly available ∼100M libraries and also comprehensive benchmark set involving molecular scores 10M library 10 diverse protein with programs, PLANTS AutoDock Vina. set, shown that, on average, more than 60 70% top 10k 1k molecules, respectively, can be retrieved while reducing number evaluations 97%, indicating robust performance protocol. With larger libraries, that proportional increase training size enhances model hits. summary, our results support use retrieve for containing hundreds millions even billions where role models becomes critical brute-force exploration such through is inaccessible reasonable time frames.

Language: Английский

---

HybridAI: Bridging the Gap Between AI Innovation and Precision Medicine (Preprint)

Published: April 4, 2025

Artificial intelligence (AI) has become a game-changing force in drug discovery, transforming target identification, lead optimization, and precision medicine. Conventional development is usually limited by excessive cost, labor-intensive experimental verification, uncertain therapeutic effects. AI-based models like AlphaFold, AtomNet, Insilico GANs have proven to be promising forecasting efficacy, toxicity, molecular interactions. However, their use still constrained inconsistency cross-therapeutic generalizability failure generalize across various disease spaces. Existing AI algorithms excel at particular tasks, protein structure prediction (AlphaFold) or virtual screening (AtomNet), but tend work isolation, limiting applicability broader contexts. The problem lies designing an system that can combine several computational approaches maximize predictive accuracy applicability. This presents HybridAI, combinational architecture integrates geometric deep learning (GDL), reinforcement (RL), federated address the shortcomings of standalone models. HybridAI bridges gaps AI-assisted discovery enhancing flexibility, robustness speedup By combining information from sources, such as ChEMBL DrugBank, make more precise predictions drug-target interactions, toxicity profiles, repurposing potential. research will (1) systematically contrast performance current models, (2) assess (3) illustrate its practical using case study on non-small cell lung cancer (NSCLC). Through bridging gap between innovation medical application, underlines power hybrid enabling personalized treatments, reducing trial-and-error inefficiencies, redefining future pharmaceutical based accelerated growth artificial calls for critical assessment validity relevance. present proposes compare predicting outcome therapy new approach, improving strength versatility. Seven including AlphaFold¹, AtomNet², GANs³, were comprehensively evaluated binding affinity four areas: oncology, antimicrobial resistance, neurodegenerative diseases, autoimmune disorders. evaluation was performed normalized metrics receiver operating characteristic (ROC-AUC), root mean square deviation (RMSD), hit-rate accuracy. novel model combines (GDL)⁴, (RL)⁵, learning⁶, validated 150 structurally diverse compound dataset derived ChEMBL⁷ DrugBank⁸. Comparative analysis indicated 78–85% target-specific design display wide variation (12–28%) generalizability. surpassed single 92% drug-kinase interactions (vs. 79% with AlphaFold¹) making 34% decrease errors compared standard ADMET predictors. cross-validated through kinase inhibitors (NSCLC correct afatinib¹⁰ 89% later confirmed vitro within time frame 14 days. results emphasize limitation individual point need architectures provide higher reliability. multi-modal methodologies, provides scalable flexible platform acceleration medicine, minimization inefficiencies development, personalization approaches.

Language: Английский

---