Ghareeb, A.E., Chang, B., Mitchener, L. et al. A multi-agent system for automating scientific discovery. Nature (2026). https://doi.org/10.1038/s41586-026-10652-y.
Get the latest international news and world events from around the world.
AI is starting to beat doctors at making correct diagnoses
Researchers show that a type of AI known as a large language model often outperformed physicians at diagnosing complex and potentially life-threatening conditions, including decreased blood flow to the heart, even in the fast-moving stages of real ER care when information is limited.
In early ER cases, the model identified the correct or a very close diagnosis in about 67% of cases, compared with roughly 50% to 55% for physicians. And the technology is only getting better.
If you walk into an emergency room (ER) in 10 years, you’ll encounter a new type of caregiver: an artificial intelligence (AI) system designed to get you a diagnosis faster and help your care team make more informed decisions. While you sit in the waiting room, you’ll be hooked up to a blood pressure cuff that’s constantly and autonomously monitored. All the while, an AI agent will be listening in while you and your doctor talk about your symptoms, ready to flag any mistakes your physician makes or suggest next steps.
This vision of AI-assisted emergency health care may soon be reality. In a new study, researchers show that a type of AI known as a large language model (LLM) often outperformed physicians at diagnosing complex and potentially life-threatening conditions, including decreased blood flow to the heart, even in the fast-moving stages of real ER care when information is limited, they report today in Science. In early ER cases, the model identified the correct or a very close diagnosis in about 67% of cases, compared with roughly 50% to 55% for physicians. And the technology is only getting better.
“Evaluating AI in medicine demands both depth and breadth across different clinical tasks and settings,” and these authors were able to incorporate both in this study, says Shreya Johri, a computer scientist at the Dana-Farber Cancer Institute who was uninvolved with the new research. Still, she notes, wide adoption of these AI systems in health care will hinge on knowing the contexts in which they’re most reliable.
Biomimetic Microfibers for Myelin-Enhancer Screening and Neural Regeneration
Roles of lysosomal small-molecule transporters in metabolism and signaling
Small-molecule transporters of the lysosomal membrane export lysosomal catabolites for reuse in cell metabolism.
These transporters often show substrate promiscuity and, conversely, a given metabolite is often exported through distinct transport routes and sometimes in different states (e.g., single amino acids versus dipeptides).
Some lysosomal transporters import metabolites into the lumen. The combination of importers and exporters can create small-molecule shuttles across the lysosomal membrane, which regulate the lumen state.
Some lysosomal transporters participate in intracellular signaling cascades. sciencenewshighlights ScienceMission https://www.cell.com/trends/cell-biology/fulltext/S0962-8924(25)00222-3 https://sciencemission.com/lysosomal-small-molecule-transporters
Remyelination requires the precise wrapping of axons by oligodendrocyte processes, a critical step for restoring neural circuit function. However, a lack of quantitative systems that recapitulate axonal geometry and chemistry has limited mechanistic and pharmacological insights into myelin wrapping. Here, we present a bioengineered microfiber platform that mimics neurite architecture and surface chemistry, enabling high-content quantification of oligodendrocyte wrapping. Through compound screening, we identified dimemorfan, a clinically used sigma-1 receptor agonist, as a potent enhancer of myelin wrapping. Dimemorfan treatment accelerated remyelination and functional recovery in demyelinated mice and promoted myelin wrapping by human induced pluripotent stem cell (iPSC)-derived oligodendrocytes.
How an HIV/AIDS tragedy spurred human evolution
Researchers show that a type of AI known as a large language model often outperformed physicians at diagnosing complex and potentially life-threatening conditions, including decreased blood flow to the heart, even in the fast-moving stages of real ER care when information is limited.
In early ER cases, the model identified the correct or a very close diagnosis in about 67% of cases, compared with roughly 50% to 55% for physicians. And the technology is only getting better.
Before antiretroviral (ARV) drugs started to become widely available in KwaZulu-Natal in 2005, there was “kind of the perfect storm,” with several unusual factors coalescing to drive a devastating epidemic, says Philip Goulder, an immunologist at the University of Oxford who led the study, which appears today in the Proceedings of the National Academy of Sciences. HIV had made few inroads into South Africa until the early 1990s, when an epidemic exploded in the heterosexual population, infecting about 40% of pregnant women in KwaZulu-Natal. (That astonishingly high prevalence persists today.) Because of a mix of genetics, limited health care, and possibly the viral subtype in circulation, infected people developed AIDS—when the destruction of the immune system threatens survival—exceptionally quickly, within about 4.5 years versus 10 years in North America.
Other studies have shown how infectious diseases, including malaria and tuberculosis, have altered the human genome. But those changes took thousands of years. “That’s what is quite exciting about this: You can see how rapidly evolution actually can occur,” Goulder says.
Similar evolutionary forces may have been at work in North America and Europe, but they are more difficult to see—and less likely to affect future generations. HIV prevalence in those regions is below 1%, and the hardest-hit group is men who have sex with men. “They are generally not a population that’s leaving behind as many offspring,” Worobey notes.
Neutrophils manufacture schizophrenia-linked protein, according to new research
The most common white blood cells in your body—immune cells called neutrophils—can make a protein nobody knew they were making, Stanford Medicine investigators have discovered. That unexpected sighting joins a growing list of hints tying schizophrenia, a disorder of the brain, to events occurring elsewhere in our bodies. The findings are summarized in a paper published in Proceedings of the National Academy of Science.
The newly noticed neutrophil nexus, as a source of the protein called C4A, links a long list of other observations that are somewhat puzzling when looked at in isolation: For example, large-scale population-genetic studies have identified C4A, already known to be produced mainly in the liver, as a pronounced risk factor in schizophrenia. People with schizophrenia tend to have increased numbers of neutrophils in their blood. And the most effective medication for schizophrenia inhibits neutrophils.
Schizophrenia affects one in every 100 persons globally almost without variation by geography or ethnicity. Its most noticeable symptoms are hallucinations, delusions and fixations. A fundamental feature of the disease is cognitive impairment: inability to think clearly, reduced working memory, disorganized thinking and behavior.
Researchers solve longstanding problem in measuring semiconductor defects
Researchers at Sandia National Laboratories and Auburn University have developed a new method to more accurately detect atomic-scale defects in electronic materials, an advance that could help improve technologies ranging from electric vehicles to high-power electronics. The study, appearing in the Journal of Applied Physics, addresses a longstanding challenge in understanding what happens at the critical boundary where a semiconductor meets an insulating layer.
At this interface, microscopic defects can trap electrical charge and quietly reduce device performance, even when the device otherwise appears to function normally. These defects can limit efficiency, increase electrical losses, and reduce the performance of advanced semiconductor devices.
Scientists commonly study these defects by comparing how a device responds to slow and fast electrical signals. However, the technique depends on knowing a key device property, the insulator capacitance, with very high accuracy. Even tiny errors can produce misleading results, sometimes making it appear that far more defects exist than are actually present.
SIRT6 protein could protect against age-related breakdown in chromatin, possibly help reverse aging
Researchers at Bar-Ilan University have successfully restored youthful patterns of DNA organization in the livers of old mice, reversing key molecular features associated with aging. The study, published in Nature Communications, identifies the protein SIRT6 as a powerful protector against age-related breakdown in chromatin, the complex system that packages DNA and controls how genes are switched on and off.
The findings suggest that aging is not simply a passive process of wear and tear, but may be driven in part by reversible changes in the way DNA is organized inside cells.
DNA inside cells is tightly folded and packaged into chromatin, a structure that acts like a biological control system for gene activity. Using advanced tools to study DNA organization and gene activity, the researchers examined multiple molecular changes in the livers of young and old mice. What they discovered was dramatic: aging disrupts chromatin architecture in the liver, causing inflammatory pathways to become overactive while weakening the metabolic programs that define healthy liver tissue.
Pronounced Neuroplasticity in the Primary Visual Cortex of the Thirteen-lined Ground Squirrel During Hibernation
Hibernating animals can show neuroplasticity throughout the hibernation season. In ground squirrels, decreased dendritic arborization in the hippocampus, somatosensory cortex, and thalamus during deep hibernation (“torpor”) suggests that this neuroplasticity is a brain-wide phenomenon. However, the degree to which neuroplasticity occurs in the visual system is not clear. While transient retinal changes have been reported during torpor, neuroplasticity beyond the retina remains unknown. Here, we characterized hibernation-related neuroplasticity in the primary visual cortex (V1), the first cortical area to receive visual information, in the thirteen-lined ground squirrel (Ictidomys tridecemlineatus). We compared neuronal morphology in Golgi-stained samples from male and female hibernating or non-hibernating squirrels. For the hibernating squirrels, brain tissue was sampled during two different epochs: torpor and inter-torpor arousal. Dendritic arborization decreased during torpor in V1 layer 2/3 pyramidal neurons, manifesting as decreases in dendritic length, number, and complexity. These changes fully reversed during inter-torpor arousal, indicating that on average dendritic arbors grew by 0.75 mm (65%) over ∼1.5 hours. No morphological differences between hibernating and non-hibernating squirrels were apparent when compared 6 months after the hibernation season. We also found no neuroplastic changes in V1 layer 4 spiny stellate neurons, unlike in this cell type the somatosensory cortex. Together, this revealed, for the first time, hibernation-related neuroplasticity in V1 in support of a brain-wide mechanism but with area-specific differences. The speed and magnitude of this naturally occurring neuroplasticity could make ground squirrel V1 a powerful translational model system for conditions requiring neuroplasticity, such as recovery from stroke.
Significance Statement This study is the first demonstration of pronounced hibernation-related neuroplasticity in the primary visual cortex of ground squirrels. Layer 2/3 pyramidal neurons in the primary visual cortex (V1) reduced arborization during torpor. Within 1.5 hours after arousal from torpor, the arborization reversed to non-hibernation levels. The extent and speed of this naturally occurring neuroplasticity could make the relatively well-understood V1 of ground squirrels a powerful translational model system. Complementing insights on neuroplasticity in V1 during development, it has the potential to be leveraged for the study of treatment mechanisms and conditions requiring neuroplasticity, ranging from neurodegeneration to recovery after stroke.
