Search

Leading biomedical researchers outline their priorities for the incoming administration.

Health has always been political, long before Donald Trump and Robert F. Kennedy Jr took power. Denying this is what’s been killing us.

Eric Topol extols a gripping account of the rise and fall of the US medical-testing company.

Eric J. Topol weighs up a book on a field sprinting to keep up with biotechnology.

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.

This Perspective describes how recent advances in artificial intelligence could be used to automate medical image interpretation to complement human expertise and empower physicians and patients.

Experimental haplotyping of whole genomes is now feasible, enabling new studies aimed at linking sequence variation to human phenotypes and disease susceptibility.

Tech giants moving into health may widen inequalities and harm research, unless people can access and share their data, warn John T. Wilbanks and Eric J. Topol.

Analyses from the US Department of Veterans Affairs databases reported residual elevated risk and health burden of long COVID at 3 years in hospitalized individuals after SARS-CoV-2 infection.

This Perspective considers the application to infectious disease modelling of AI systems that combine machine learning, computational statistics, information retrieval and data science.

The antiplatelet drug clopidogrel helps prevent stent-associated thrombosis, but the antiplatelet effects are quite variable and the clinical consequences can be serious. New findings show that the variability in clopidogrel efficacy is affected by the enzyme paraoxonase-1 (PON1), which is required for clopidogrel bioactivation (pages 110–116).

Much debate surrounds the utility of CYP2C19^*2 genotyping in patients receiving clopidogrel after coronary artery stenting. The effectiveness of its use in a point-of-care setting has now been examined and, given the substantial incremental suppression of platelet reactivity achieved, its routine use might soon be a reality.

A smartphone app that combines smartwatch and activity tracker data together with self-reported symptoms allows continuous monitoring of SARS-CoV-2 infection.

An outbreak of polio in 1950s Denmark led one hospital to pioneer mechanical ventilation, constant monitoring of vital signs and other innovations that are saving lives to this day.

Large language model-based agentic systems can process input information, plan and decide, recall and reflect, interact and collaborate, leverage various tools and act. This opens up a wealth of opportunities within medicine and healthcare, ranging from clinical workflow automation to multi-agent-aided diagnosis.

Induced pluripotent stem cells generated from older individuals have higher levels of epigenetic and genetic abnormalities.

RETFound, a foundation model for retinal images that learns generalizable representations from unlabelled images, is trained on 1.6 million unlabelled images by self-supervised learning and then adapted to disease detection tasks with explicit labels.

This Opinion article argues that capturing phase information in human genomics studies is crucial for important aims such as understanding how genotype contributes to phenotypes. existing approaches for phasing will need to be improved in order to meet this goal.

Germany and Taiwan top a ranking drawn up pre-crisis by architect of US Affordable Care Act; United States trails behind.

Whole-genome sequence analysis identifies five independent risk loci for Lewy body dementia and demonstrates overlapping genetic architecture with Alzheimer’s and Parkinson’s diseases.

Organ damage is often detected late, when treatment options are limited. The use of artificial intelligence to continuously monitor a patient’s medical data can identify people at risk of imminent kidney injury.

COVID’s evolution signals the importance of rational vaccine design based on broadly neutralizing antibodies.

Here, by sub-kb Hi-C and chromosome engineering, the authors visualize bacterial transcriptional units, showing that they form transcription-induced domains. Transcription-induced domains enforce constraints on nearby sequences, affecting their localization and dynamics.

Mechanical stress associated with cardiac fibrosis, modelled by a hydrogel with controllable stiffness, affects regulation of the phenotypes of human cardiomyocytes by a portion of the long noncoding RNA ANRIL.

The possibility of deep population resequencing of genes has generated excitement over its potentially promising role in understanding complex human traits. A new study has now demonstrated the utility of this approach, reporting the resequencing of a lipid metabolism gene in a large multiethnic population and definitively showing that coding variants in the gene are associated with plasma triglyceride levels.

In many sequencing applications, it is sufficient to sequence selected portions of a genome rather than the complete genome. Tewhey et al. describe an approach for massively parallel genome targeting that relies on PCR in microdroplets generated by a microfluidic device.

Long COVID is an often debilitating illness of severe symptoms that can develop during or following COVID-19. In this Review, Davis, McCorkell, Vogel and Topol explore our knowledge of long COVID and highlight key findings, including potential mechanisms, the overlap with other conditions and potential treatments. They also discuss challenges and recommendations for long COVID research and care.

Lipid dynamics are analyzed in a model for the mitochondrial inner membrane using coarse-grained molecular dynamics. The model displays biological scales of curvature where strong partitioning of cardiolipin to negative curvature regions is observed.

The CommonMind Consortium sequenced RNA from dorsolateral prefrontal cortex of subjects with schizophrenia (N = 258) and control subjects (N = 279), creating a resource of gene expression and its genetic regulation. Using this resource, they found that ∼20% of schizophrenia loci have variants that may contribute to altered gene expression and liability.

This review discusses generalist medical artificial intelligence, identifying potential applications and setting out specific technical capabilities and training datasets necessary to enable them, as well as highlighting challenges to its implementation.

A wearable ultrasound patch enables the continuous monitoring of cardiovascular performance outside the intensive care unit.

This Review outlines the current state of scientific evidence on long COVID, discusses its impacts on patients, health systems, economies and global health metrics, and proposes a forward-looking research and policy roadmap.

Patrick Ellinor and colleagues report a genome-wide association study identifying variants in KCNN3 associated to lone atrial fibrillation.

This Review discusses the advantages and limitations of self-supervised methods and models for use in medicine and healthcare, and the challenges in collecting unbiased data for their training.

Multimodal artificial intelligence models could unlock many exciting applications in health and medicine; this Review outlines the most promising uses and the technical pitfalls to avoid.

With only a limited number of clinical trials of artificial intelligence in medicine thus far, the first guidelines for protocols and reporting arrive at an opportune time. Better protocol design, along with consistent and complete data presentation, will greatly facilitate interpretation and validation of these trials, and will help the field to move forward.