Toggle light / dark theme

Get the latest international news and world events from around the world.

Log in for authorized contributors

Macronutrient composition and genetic background determine the response to a ketogenic diet in mice

Zhang et al. show that dietary protein and genetic background shape ketogenic diet response. Unlike other strains, C57BL/6J mice eat less and lose weight on a low-protein KD. NNT and NLRP12 mutations, in combination with the diet, activate a stress and inflammation response to induce GDF15 and LCN2, which reduce food intake and decrease weight.

Human-protein CAR-T design avoids anti-drug antibodies and extends persistence

A first-in-human study by Mass General Brigham Cancer Institute investigators found that a chimeric antigen receptor (CAR) T-cell therapy called TriPRIL reduced or eliminated signs of cancer in 12 of 15 patients with relapsed or refractory multiple myeloma. TriPRIL — short for trimeric APRIL (a proliferation-inducing ligand) — recognizes two proteins on myeloma cells instead of just one. Researchers published their findings in The Lancet.

Multiple myeloma is a cancer of plasma cells, which are immune cells that make antibodies. To create CAR T-cell therapy, scientists modify a patient’s T cells so they can find and attack myeloma cells. Existing CAR T-cell therapies recognize B-cell maturation antigen (BCMA) on myeloma cells. TriPRIL was engineered by Mass General Brigham researchers to make the CAR-T cells recognize both BCMA and a second protein, transmembrane activator and CAML interactor (TACI). Because the treatment is based on human proteins, the researchers believe it may be less likely to lead to anti-drug antibodies that can reduce the long-term persistence and effects of other CAR-T cells.

“BCMA-targeted CAR T-cell therapy has transformed multiple myeloma treatment, but most patients eventually relapse,” said co-senior author Marcela V. Maus, MD, PhD, director of the Mass General Brigham Cancer Institute’s Cellular Immunotherapy Program. “TriPRIL adds TACI as a second target, allowing it to find and attack myeloma cells, even if they stop expressing BCMA.”

Zero Trust Is the Answer to Machine-Speed Agentic AI Attacks

Chuck Brooks is the president of Brooks Consulting International and one of Executive Mosaic’s GovCon Experts.

The cyber arms race has moved into a new stage. These days, data exploitation and exfiltration happen at speeds that are unmatched by conventional security operations. Patching cycles, which used to provide a significant window of protection, are losing their operational significance. Businesses that continue to rely on human-paced, reactive defenses are already falling behind.

Autonomous, machine-paced systems are increasingly controlling devices that were previously controlled by proficient human adversaries. Attack timelines have been reduced from days or weeks to minutes or seconds thanks to agentic AI—self-directed agents that can perform reconnaissance, recursive exploitation, lateral movement, adaptation and persistence with little human supervision.

“Clean and Conscious” Brian Clement of the Hippocrates Institute and Bill Faloon — Nov 19th

Brian will give a presentation on “Clean and Conscious” Followed by an Age Reversal Update by Bill Faloon After the presentation there will be a book signing by Brian Clement of his book “Quantum Human”, and join us for a banquet in honor of our speakers. Doors open at 6:00 PM and the main event begins at 7:00 PM, Thursday, Nov. 19th.

Language model teams as distributed systems

Poorly coordinated AI teams compound errors, waste resources, and underperform single agents. We draw on decades of distributed computing to ask: When do teams help? How many agents do you need? And how does structure impact performance?

💡 The Big Takeaway:

It turns out the secret to powerful AI isn’t just deploying *more* agents—it’s designing smarter systems for them to collaborate.

🔗 Read the full 🤖 More AI isn’t always better.

We’ve all heard the hype: “Just put multiple AI agents in a room and let them figure it out!” But a fascinating new study published in the Proceedings of the National Academy of Sciences (PNAS) reveals a major catch: poorly coordinated AI teams actually compound errors, waste resources, and underperform a single AI agent. 📉

Researchers drew on decades of knowledge from distributed computing to ask three big questions about AI teams:

1️⃣ When do they actually help? 2️⃣ How many agents do you really need? 3️⃣ How does team structure impact performance?

Marcus Du Sautoy

AI can see what humans can’t.

Marcus du Sautoy, Professor of Mathematics at the University of Oxford, admits that just three months earlier, he thought AI was essentially useless for the kind of mathematics he was doing.

Then he tried it again.

What happened next fundamentally changed his perspective.

He had been wrestling with a mathematical conjecture given to him by his PhD supervisor nearly 40 years ago. The conjecture concerned a symmetrical mathematical structure that, according to his supervisor, would look the same regardless of where or from which direction one examined it.

For 20 years, du Sautoy tried to prove the conjecture was true. At one point, he thought he had succeeded, only to discover a mistake in his proof.

He then spent another 20 years trying to prove the opposite — that the conjecture was false. But he kept hitting a wall. He simply couldn’t find a way to make the interesting mathematical properties of the structure emerge.

Brain organoids: Experimental access to human-specific brain development

Human brain development is largely inaccessible to mechanistic investigation, and evolutionarily divergent programs cannot be fully modeled in other species. Here, Kastli et al. review brain organoids as powerful experimental systems to unravel the mechanisms of human neurodevelopment and its neurodivergences.

Cancer cell plasticity: new insights from developmental programs

Tumors represent a heterogeneous set of neoplastic diseases, each composed of an intricate network of cancer cells residing in multiple alternative phenotypic states. Transitions between these phenotypic states, often termed ‘phenotypic plasticity’, enable them to execute specific steps in tumor progression and to develop therapeutic resistance. The phenotypic plasticity of tumor cells is mediated, in part, by cellular processes that orchestrate normal embryonic development and are hijacked by tumors. In this review, we discuss the contributions of these developmental programs to cancer cell phenotypic plasticity. We focus on epithelial–mesenchymal transition and ciliogenesis programs and discuss new insights into the mechanistic roles of these cellular processes in cancer progression and response to treatment.

Home-based auricular vagus nerve stimulation shows early promise for kids and teens with migraines

As clinicians and researchers, we often see how heavily migraine attacks weigh on children and adolescents, yet non-pharmacological, child-friendly treatment options remain strikingly scarce. For years, our team at Nanfang Hospital and collaborating institutes have been looking into ways to bridge this gap.

Recently, our team set out to explore whether a gentle, at-home wearable technology—transcutaneous auricular vagus nerve stimulation (taVNS)—could offer a safe and feasible way to help young patients manage acute migraine episodes.