Medicine & Health

This work builds BoneGraph, a bone-science-specific system delivered as a five-tab web application over a shared substrate of 7,449 documents, 248,629 SPECTER2 passage vectors, and a bone knowledge graph of 1,597 concepts and 1,699 causal relations, reporting MRR 0.928 on a 30-question seven-domain retrieval benchmark, 92.6% accuracy for the bone-region classifier on held-out MURA, and an increase in answer accuracy from 42% to 78% when the correct passage is supplied, with all inference performed locally on a single NVIDIA Jetson AGX Orin and no third-party API calls.

The work introduces LDDM (Large Drug Discovery Model), a unified 3D generative framework supporting constrained and unconstrained docking, fragment linking and growing, and de novo design, together with a programmable design algorithm that produces synthetically accessible compounds satisfying fine-grained objectives; the authors experimentally validated designed or optimised ligands for five therapeutically relevant protein targets, achieving high success rates and identifying molecules with confirmed binding affinity while synthesizing only a small number of generated compounds, with the best designs structurally characterised by NMR spectroscopy and X-ray crystallography indicating high prediction accuracy.

This review systematically surveys recent advances in natural biodegradable polymer-based microneedle drug delivery systems in terms of material selection, fabrication strategies, and controlled release mechanisms, and introduces a rational design framework that systematically integrates therapeutic objectives, polymer properties, mechanical performance, and release kinetics, while discussing translational challenges such as mechanical limitations, manufacturing scalability, stability concerns, and regulatory considerations, as well as emerging trends including stimuli-responsive systems, nanocarrier integration, personalized drug delivery, and data-driven strategies such as artificial intelligence and machine learning, arguing for a shift from empirical formulation toward a predictive, en

The study proposes RSAUNet, a deep learning architecture that combines residual convolutional blocks, Swin Transformer blocks, and attention mechanisms inside a U-Net backbone for MRI prostate cancer segmentation, reporting a Dice coefficient of 0.998 and a Jaccard index (IoU) of 0.965 on a public Kaggle prostate annotation dataset, together with an ablation study tracking loss, accuracy, Dice, and mean IoU as each component is added.

This work presents LLMsFold, a computational framework that identifies binding pockets geometrically, has Llama-3.3-70B generate candidate small molecules as SMILES strings, evaluates each with the Boltz-2 co-folding model for bound pose and binding affinity, and iteratively refines candidates through a feedback loop, yielding molecules for ACVR1 and CD19 that pass drug-likeness, synthetic accessibility, and novelty filters, with the ACVR1 candidate reaching a predicted affinity probability of 0.953 and predicted pIC50 of about 10.72 and the CD19 Pocket 1 candidate reaching a predicted pIC50 of about 7.73.

This work introduces latent generative search for binder design, a framework that uses reward-guided search at inference time to steer the Proteina-Complexa generative model, which codesigns sequence and structure together in a continuous latent space and thereby removes the inverse-folding step; in a screen of more than one million designs by multiplexed phage display, it produced more validated binders than every other method tested, its codesigned sequences surpassed post hoc redesign, it delivered high-affinity binders across therapeutic receptors, a viral attachment protein and intracellular signalling targets, and it generated the first de novo proteins that bind a free carbohydrate, including one that discriminates between blood-group antigens.

Using tissue staining and RNA sequencing of mouse embryos 7.5 days after conception, red fluorescent lineage tracing, directed differentiation of human pluripotent stem cells, and a search across monkeys, chickens, zebrafish and even acorn worms, this work proposes that the brain is not made by a single type of starter cell but by two non-mixing progenitor populations, one forming the hindbrain and the other the forebrain and midbrain, and it establishes an efficient way to coax stem cells into hindbrain motor neurons.

A man with a slowly progressing form of motor neuron disease (ALS) caused by a rare CHCHD10 mutation became the first person to receive an RNA antisense oligonucleotide therapy targeting his specific disease-causing mutation; after three 50-milligram and three 75-milligram doses delivered into his spine between April 2024 and April 2025, he had no serious side effects, and one year later his blood neurofilament light chain levels had fallen to the normal reference range, his motor, breathing and neurological function scores had improved, breathing and cognition scores remained stable, and he continued to work as a physician.

This review, based on a structured literature search of PubMed/MEDLINE, Web of Science and Scopus (2000-2026) supplemented by guidance documents from OECD, FDA, EMA, EFSA and EPA, traces the development of AI in toxicology from rule-based expert systems to deep learning, large language model and multimodal architectures, reports that graph neural networks and multi-task deep learning have shown competitive performance in selected benchmark studies of drug-induced liver injury, hERG cardiotoxicity and Ames mutagenicity, and concludes that inconsistent external validation, limited generalisation of endpoint-specific models and hallucination in large language models leave unresolved regulatory risks, so AI models augment but cannot yet replace experimental toxicology.