Toggle light / dark theme

AI fails classic attention test, with longer word lists triggering dramatic accuracy collapse

Giving AI a classic psychological test reveals an inherent weakness in LLM decision-making abilities. Suketu Patel and colleagues explored how transformer-based machine attention differs from human attention by testing AI models on the “Stroop task,” in which words for colors are printed in colored ink, and participants are asked to name the ink color of each word while ignoring its meaning.

The findings are published in the journal PNAS Nexus.

The task is clinically used to assess executive control, especially a person’s ability to inhibit an automatic response. Although humans generally take longer to answer correctly when words and colors are mismatched than when they match, they can still perform stably and with high accuracy even on long word lists.

RNA-guided transposon mechanics show use of figure-eight intermediate and direct-transfer route

IS110 transposons are a large, diverse family of bacterial insertion sequences (IS elements)—small, mobile DNA elements that can move from one genomic location to another. They have recently attracted broad interest due to the finding that some of these transposons use a bridge RNA (bRNA) to recognize both donor DNA and target DNA.

Upon this discovery, researchers hoped that bRNA-guided transposon systems could offer a genome-editing strategy distinct from classical CRISPR-Cas nucleases and thereby enable programmable DNA integration. However, it remained unclear how IS110 elements insert donor DNA into target sites and whether these elements rely on one or multiple reaction pathways.

Now, a new study led by Xue Chaoyou from the Tianjin Institute of Industrial Biotechnology of the Chinese Academy of Sciences, in collaboration with Lou Huiqiang at China Agricultural University and RAO Shuquan from the Chinese Academy of Medical Sciences, answers these questions by showing that RNA-guided IS110 transposons use two mechanistically distinct pathways to mobilize DNA.

Early immunotherapy aids in treating potentially fatal fungal pneumonias in preclinical models

A new study led by researchers at The University of Texas MD Anderson Cancer Center has shown that early administration of immunotherapy with standard antifungal treatment improved outcomes and largely alleviated immune system paralysis caused by fungal lung infections in preclinical models. These findings could herald new clinically relevant strategies for treating a variety of life-threatening invasive fungal pneumonias, which disproportionately affect immunocompromised cancer patients.

The study, published in the Proceedings of the National Academy of Sciences, was led by Sebastian Wurster, M.D., assistant professor, and Dimitrios P. Kontoyiannis, M.D., Ph.D., professor, both of Infectious Diseases, Infection Control and Employee Health.

“Despite an expanded arsenal of antifungal treatments, immune system dysfunction is still a major cause of failure when treating infections, with significantly high morbidity and mortality rates associated with pneumonias caused by opportunistic molds. There is an urgent need for adjunct immune-enhancing therapies to improve outcomes,” Kontoyiannis said. “Our research shows that adding an immune checkpoint inhibitor to antifungal treatments is helpful in experimental mold pneumonias, especially when given early.”

Predictors of first-year statin medication discontinuation: A cohort study

The discontinuation of statin medication is associated with an increased risk of cardiovascular and cerebrovascular events and, among high-risk patients, all-cause mortality, but the reasons for discontinuation among statin initiators in clinical practice are poorly understood.

Embryonic transplantation and ischemic memory deficit

Transient forebrain ischemia is associated with selective neuronal vulnerability and persistent memory deficit. This study compares functional outcome and morphological changes in rats subjected to post-ischemic CA1 or hilus/dentate gyrus region hippocampal fetal transplantation. Ischemia was produced by bilateral common carotid artery occlusion with hypotension. Fetal hippocampal neurons were transplanted into both sides of the CA1 or hilus/dentate gyrus region of the dorsal hippocampus, 1 week post-ischemia. Four weeks post transplantation, the rats underwent behavioral testing for 5 consecutive days using the water maze trial. All animals were perfusion fixed for morphological studies. Transplants in the CA1 region of the dorsal hippocampus were associated with memory and morphological recovery, while grafts placed into the hilus/dentate gyrus region of the dorsal hippocampus were not. Similarly, neurons transplanted in the CA1 region of the dorsal hippocampus were morphologically similar to CA1 pyramidal cell neurons and stained positive with calbindin D(28k). In contrast the grafts transplanted into the hilus/dentate gyrus region of the dorsal hippocampus were morphologically heterogeneous and staining with calbindin D(28k) was not as robust. Post-ischemic transplantation in the CA1 region of the dorsal hippocampus is effective in improving memory and morphological function.

PubMed Disclaimer

Severe obesity in human HFpEF alters contractile protein function and organization

Heart failure with preserved ejection fraction (HFpEF) causes substantial morbidity and mortality and has few effective therapies. Its phenotype has changed over time, with morbid obesity and metabolic defects supplanting hypertension and cardiac hypertrophy. We reveal that cardiomyocytes from patients with severe obesity and HFpEF have very depressed contractile reserve, including reduced calcium-and length-stimulated tension, power, and myosin activation compared with less-obese HFpEF and nonfailing (NF) controls with or without obesity but similar to those with advanced HF and reduced ejection fraction. Myocyte defects correlate with body mass index and exercise hemodynamics in patients with HFpEF but not NF and appear reversible upon weight loss. Increased troponin I phosphorylation at threonine 181 occurs only in heart failure with obesity, contributing to sarcomere dysfunction.

An ensemble pipeline, PhageHost, for phage tail fiber discovery and accurate Klebsiella pneumoniae host prediction using protein language models

Wu et al. present an ensemble pipeline, PhageHost, comprising a protein language model, TailSeek, for tail fiber detection from phage and prophage genomes and a deep learning model, HostBuster, that integrates tail fiber features with host information to predict the lytic potential of phage–K. pneumoniae pairs.

AI ‘super-antigen’ vaccine could protect against whole families of viruses

A groundbreaking new vaccine technology using artificial intelligence could offer immunity against entire families of viruses and protect against future mutations with a single injection.

Researchers say this could prevent future pandemics before they emerge, saving millions of lives and sparing countries from the necessity of lockdowns.

A “super-antigen” has been developed through AI machine learning that meticulously analyses past and current outbreaks to pinpoint the essential elements for the survival of viruses.

Why energy fades with age: Missing membrane lipid may destabilize mitochondria

Why do cells age—and why do we lose our energy and vitality as we get older? This question is one of the central challenges of modern biomedicine. The focus is particularly on mitochondria—tiny cellular organelles long known as the cell’s powerhouses but now understood as dynamic control centers that not only produce energy, but also coordinate cellular communication, adaptation, and many of the processes essential for life.

They supply us with the energy that our body needs for movement, growth, and repair processes. But as we age, these powerhouses begin to slow down. It has long been known that their function declines with age. But until now, the mechanisms driving this gradual decline have been poorly understood.

Focus on membrane lipids For a long time, it was assumed that genetic damage within the mitochondria themselves was primarily responsible. A study now published in Nature Communications by an international research team led by Dr. Maria Ermolaeva of the Leibniz Institute on Aging—Fritz Lipmann Institute (FLI) in Jena provides a surprising answer to this question: A key factor appears to be the imbalance in the structure of the mitochondrial network, which is caused by the absence of a major lipid in the membrane composition.

/* */