News & Comment

Decentralized yet coordinated networks of specialized artificial intelligence agents, multi-agent systems for healthcare (MASH), that excel in performing tasks in an assistive or autonomous manner within specific clinical and operational domains are likely to become the next paradigm in medical artificial intelligence.

Antigen design and optimization for broader yet targeted immune stimulation, and targeted antigen delivery, are reshaping vaccine development for infectious diseases and cancer.

Advances in base editing and prime editing, coupled with optimized delivery strategies, are enhancing the efficiency, safety and versatility of genome editing for the treatment of a widening range of genetic diseases.

Combining neuroimaging with molecular and cellular data reveals how microscale features shape large-scale brain networks.

Cas9-based ribonucleoproteins engineered via directed evolution and encapsulated in optimized lipid nanoparticles improve genome-editing efficacies in the livers and lungs of mice.

High-throughput screening led to the identification of artesunate as an antifibrotic agent that targets the glycoprotein myeloid differentiation factor 2 in the infarcted heart.

Organic nanoprobes facilitate the imaging of small deep-seated tumours in mice via the X-ray-mediated generation of long-lasting luminescence.

Naming brain–computer interfaces according to their intended application will assist stakeholders in the evaluation of the benefits and risks of neurotechnologies.

An autologous transplant of islets derived from chemically induced pluripotent stem cells allowed an individual with type 1 diabetes to forgo insulin.

Genome editing is powering the generation of increasingly safer and potent multifunctional therapeutic cells.

An agonist of the neurokinin 2 receptor induced weight loss and increased energy expenditure in obese mice and monkeys with diabetes.

Foundation models can be advantageously harnessed to estimate missing data in multimodal biomedical datasets and to generate realistic synthetic samples.

RNA-targeting CRISPR screens reveal hundreds of functional long non-coding RNAs that are crucial for cell survival and implicated in cancer progression.

Virus-like particles can be evolved to enhance the delivery of ribonucleoproteins and to increase particle production.

A large vision–language model trained to generate chest X-ray reports shows promise as an assistive tool for radiologists, particularly for typical cases in outpatient settings.

More off-the-beaten-track strategies are needed for enhancing the accumulation of drugs at target sites, for improving the drugs’ cell-type specificity and for reducing their off-target toxicity.

Proteins that tag surface receptors for degradation by triggering their endocytosis can be computationally designed so that they do not compete with native receptor ligands for binding.

DNA microbeads loaded with an agonist for Wnt and injected into retinal organoids allow for the spatiotemporal control of gradients of the morphogen to better direct organoid development and maturation.

A biomarker in retinal images identified by a deep-learning algorithm can predict biological age as well as mortality and morbidity.