Science & Technology News Review | November 18, 2026

RISOLNET • KNOWLEDGE ENGINE

Science & Technology News Review

November 18, 2026 | Five Stories | Editorial cutoff: 08:00 CET

This edition examines five developments across Earth observation, neuroscience, artificial intelligence and materials engineering. From lightning photographed above the South China Sea to proteins selectively transported into the brain, the stories reveal how new methods of observation can challenge established assumptions about complex systems.

The research also raises questions about practical implementation. Machine learning can improve structural design while retaining human judgment, but it can also be influenced by pharmaceutical advertising. Meanwhile, experiments with network-inspired materials suggest that deliberately introducing structural disorder may sometimes improve mechanical performance.

The selected items were originally reported between November 11 and November 17, 2026. They include peer-reviewed research and an institutional observational report. Publication dates, interpretations and limitations are distinguished throughout.

Space & Earth Observation
01

Lightning Above the South China Sea Reveals the Complexity of Tropical Storms

An astronaut’s nighttime photograph captures intense thunderstorms alongside the artificial lights of cities, fishing fleets and offshore infrastructure.

NASA Earth Observatory published a remarkable orbital photograph on November 17, 2026, showing thunderstorms over the South China Sea near Borneo. The photograph was acquired on October 28 by an astronaut aboard the International Space Station using a Nikon Z9 camera.

The image captures several distinct phenomena. Lightning illuminates the upper portions of large thunderstorm clouds, while the lights of Singapore and neighboring islands appear along the horizon of the photographed region. Clusters of blue lights offshore are likely associated with fishing vessels, while scattered orange lights may correspond to gas flares.

Of particular meteorological interest are the overshooting cloud tops. These structures occur when powerful thunderstorm updrafts push cloud material above the surrounding anvil, sometimes reaching into the lower stratosphere. Their presence can indicate vigorous atmospheric convection.

Orbital photography provides a perspective unavailable to observers on the ground. Lightning that appears as a brief flash beneath a cloud deck can illuminate extensive regions when viewed from above. This makes astronaut photographs useful contextual observations alongside satellite instruments specifically designed to measure lightning and cloud properties.

NASA estimates that approximately 2,000 thunderstorms are active worldwide at any given moment, producing nearly 60 lightning flashes per second. These are broad global estimates rather than measurements derived from this individual photograph.

The image is an observational record, not a new peer-reviewed study. Its scientific value lies in the combination of visible storm structure, lightning illumination and human activity within a single nighttime scene.

Lightning and luminous electrical phenomena photographed above thunderstorms from orbit
Contextual photograph of upper-atmospheric electrical activity, not the October 28, 2026 observation. Source: EarthSky, 2017.

Analysis

Evidence: NASA’s October 28 astronaut photograph documents illuminated thunderstorm clouds, overshooting tops and artificial lights over the South China Sea.

Interpretation: Orbital photography can provide useful visual context for understanding atmospheric convection and the geographical distribution of human activity.

Limitations: A single photograph cannot establish lightning frequency, storm intensity or long-term meteorological trends. Some artificial light sources are identified probabilistically.

Event: Photograph acquired October 28; NASA Earth Observatory article published November 17, 2026.

Neuroscience & Life Science
02

A Protein Atlas Reveals How the Blood-Brain Barrier Supplies the Brain

Experiments in mice identify a selective molecular transport system that may become important for understanding neurodegenerative disease.

The blood-brain barrier is usually described as a protective interface separating the brain from potentially harmful substances circulating in the bloodstream. Research from Gladstone Institutes suggests that its function is more sophisticated: the barrier also actively imports selected proteins from other organs.

In research reported on November 11 and published in Cell, a team led by Andrew Yang developed a method for tracking proteins moving from the bloodstream into mouse brain tissue.

The scientists used protein-labeling chemistry to follow 861 circulating proteins. Of these, 92 were detected crossing into the brain. Many originated in organs outside the nervous system, particularly the liver.

The results indicate that the barrier is not simply a passive filter. It uses selective transport mechanisms to import particular molecules, which may then be delivered to specific populations of brain cells.

One example is apolipoprotein A1, or APOA1, a protein associated with lipid transport and produced largely outside the brain. The researchers found evidence that its delivery supports astrocytes and pathways involved in maintaining neuronal connections.

The team also investigated mouse models of Alzheimer’s disease. Before the barrier displayed more obvious physical leakage, its ability to import certain protective proteins had already deteriorated.

In a further experiment, mice engineered to produce additional APOA1 in the liver showed fewer disease-associated abnormalities. This result suggests that communication between peripheral organs and the brain could become an important therapeutic research direction.

However, the experiments were performed in mice. They do not demonstrate that increasing APOA1 in humans prevents Alzheimer’s disease or that the identified transport pathways can already be exploited safely for clinical treatment.

Conceptual illustration of interconnected brain regions and neural activity
Contextual neuroscience illustration accompanying a separate Stanford Medicine study. It does not depict the Gladstone protein-transport experiments. Source: Stanford Medicine.

Analysis

Evidence: The researchers identified 92 circulating proteins transported into mouse brains and observed changes in selected transport pathways in Alzheimer’s disease models.

Interpretation: Brain health may depend partly on selective molecular communication with peripheral organs, providing potential targets for future drug-delivery and neuroprotection research.

Limitations: Mouse findings cannot establish therapeutic effectiveness in humans. The atlas is an initial resource, not a complete map of human blood-brain barrier transport.

Publication: Peer-reviewed research in Cell, reported November 11, 2026.

Artificial Intelligence & Engineering
03

An AI Copilot Brings Human Judgment Into Structural Optimization

MIT researchers demonstrate that designers can influence an algorithm while it works, improving manufacturability without a major computational penalty.

Topology optimization allows engineers to design structures by distributing material according to mechanical performance requirements. The resulting geometries can be lightweight and strong, but they are sometimes difficult to manufacture or inspect.

Researchers at MIT have developed a machine-learning-assisted approach that introduces human feedback during the optimization process rather than waiting until the final structure has been generated.

The work, described in Communications Engineering and reported by MIT on November 13, uses a trained model to identify regions of a developing structure where a human designer is likely to request modifications.

The model was trained using synthetic examples of design feedback addressing criteria such as structural performance, appearance and manufacturability. During optimization, the system presents selected regions for intervention.

In the current implementation, users can make designated regions thicker or thinner. The optimization process then incorporates those decisions while continuing to generate the design.

In the researchers’ demonstrations, the collaborative method improved manufacturability and linear buckling load by 39 percent under the reported evaluation conditions. It added approximately 15 seconds to the total design time compared with a fully algorithmic approach.

This is particularly relevant to civil engineering, where an optimized structure must satisfy not only mathematical performance criteria but also practical construction requirements, safety standards and professional judgment.

The researchers do not claim that the system replaces structural engineers. Its purpose is to combine computational optimization with human expertise, potentially reducing the gap between theoretically efficient designs and structures that can actually be built.

Conceptual image combining engineering blueprints and an artificial intelligence design interface
Contextual MIT illustration from a separate engineering software article, not the structural optimization experiment. Source: MIT News.

Analysis

Evidence: The research demonstrated an interactive optimization method with improved reported structural performance and limited additional computation time.

Interpretation: Human-in-the-loop AI may be particularly valuable in engineering fields where manufacturability and professional accountability are as important as numerical optimization.

Limitations: The present interface supports only restricted geometric interventions. The reported improvements do not establish performance across all engineering applications or construction environments.

Publication: Peer-reviewed research in Communications Engineering; MIT institutional communication dated November 13, 2026.

Artificial Intelligence & Digital Health
04

Pharmaceutical Advertising Can Shift AI Treatment Recommendations

A controlled study reveals that commercially framed information can influence medical language models even when their answers remain superficially reasonable.

Large language models are increasingly used to answer health questions, summarize clinical information and compare treatment options. Their ability to process surrounding text also creates a potential vulnerability: irrelevant or commercially motivated content may influence the recommendations they produce.

Researchers at the Icahn School of Medicine at Mount Sinai investigated this issue in a study published online in The Lancet Digital Health on November 11, 2026.

The researchers evaluated 12 AI models using 33 physician-written clinical scenarios. Each scenario was presented either without advertising or with a clearly labeled pharmaceutical advertisement placed immediately before the clinical question.

The clinical information and available treatment choices remained unchanged. The experimental variable was the presence of the advertisement.

Across 258,559 model responses, the advertised drug was selected in 47.6 percent of responses when the advertisement was present, compared with 35 percent when it was absent.

In four clinical scenarios, the researchers observed an especially large shift: models that had not selected the advertised drug without the advertisement selected it in every response when the advertisement was introduced.

The results are significant because the generated recommendations could still appear medically reasonable. Commercial influence may therefore be difficult to detect simply by checking whether an answer contains an obviously incorrect treatment.

The study did not establish that advertisements caused models to recommend unsafe drugs in general. It also did not measure changes in real-world prescriptions, patient behavior or health outcomes.

The findings support evaluating medical AI systems under realistic conditions that include promotional material, conflicting information and other potentially distracting inputs.

Editorial graphic discussing the vulnerability of medical artificial intelligence systems to misleading information
Contextual graphic about a different Mount Sinai study of medical AI misinformation, published in February 2026. Not a figure from the pharmaceutical advertising experiment. Source: HLTH.

Analysis

Evidence: A controlled comparison found that adding a pharmaceutical advertisement changed the frequency with which tested AI models selected the advertised treatment.

Interpretation: Medical AI evaluations should examine not only whether recommendations are clinically plausible, but also whether they are influenced by irrelevant commercial information.

Limitations: The experiment does not establish real-world patient harm or measure prescribing outcomes. Results depend on the selected models, scenarios and advertising format.

Publication: Peer-reviewed controlled study, The Lancet Digital Health, November 11, 2026.

Technology & Materials Science
05

Nature-Inspired Networks Produce More Resilient Metamaterials

Experiments show that structures built from beams of different sizes can distribute damage more effectively than conventional repeating lattices.

Engineered metamaterials obtain many of their unusual mechanical properties from carefully designed internal structures. Conventional approaches often use repeating unit cells, creating regular patterns that can be optimized for stiffness, strength or energy absorption.

Researchers from MIT and Northeastern University have demonstrated a different strategy inspired by complex networks found in nature, including leaf veins and insect wings.

In research published in Nature and announced on November 11, the team introduced scale-rich metamaterials. Rather than relying on identical repeating units, these materials contain interconnected beams of different lengths and thicknesses.

The researchers developed an algorithm that generates networks through progressively smaller intersecting structural elements. By adjusting the initial beam thickness and how quickly subsequent beams become thinner, they could generate materials with different mechanical characteristics.

The team then manufactured polymer samples using three-dimensional printing. Individual samples contained interconnected features ranging from micrometers to centimeters in size.

Mechanical tests included compression and impact loading. Compared with conventional lattice and honeycomb structures, the network-inspired samples displayed improved resilience under the evaluated conditions.

A key advantage appears to be the distribution of damage. In a regular lattice, failure of certain structural elements can produce concentrated weaknesses. A network containing features at multiple scales may offer alternative paths for distributing mechanical forces.

The researchers also demonstrated a protective structure inspired by the thick rind of a pomelo. By varying network density, the material could cushion a fragile object placed within it.

Potential applications include protective equipment, impact-resistant packaging and acoustic structures. These remain possible future uses rather than products validated for commercial deployment.

Researchers examining a small three-dimensional printed metamaterial sample in a laboratory
Contextual photograph from a separate MIT project on graph-based metamaterial design. Not the scale-rich network samples reported in November 2026. Source: MIT Morningside Academy for Design.

Analysis

Evidence: Researchers fabricated network-inspired polymer metamaterials and tested their response to compression and impact, observing improved resilience relative to selected conventional structures.

Interpretation: Structural disorder and variation across multiple length scales can sometimes improve mechanical performance rather than undermine it.

Limitations: Laboratory-scale polymer samples do not establish the durability, manufacturing economics or certification performance of finished protective products.

Publication: Peer-reviewed research in Nature, November 11, 2026.

The RKE Signal

The five developments in this edition share a common theme: complex systems often become more understandable when researchers examine the interactions between their individual components rather than treating the entire system as a single object.

NASA’s orbital photograph shows atmospheric phenomena alongside human activity, demonstrating the value of observing familiar processes from an unfamiliar perspective. The image cannot replace dedicated meteorological measurements, but it provides a compelling view of how storm dynamics and artificial illumination coexist across a large geographical region.

The Gladstone research challenges the idea of the blood-brain barrier as a simple protective wall. Its selective transport mechanisms suggest that communication between peripheral organs and the nervous system may be essential to brain health. Understanding this relationship could eventually support new therapeutic strategies, although the present evidence comes from mouse experiments.

The two artificial-intelligence stories expose different aspects of human-machine interaction. MIT’s structural optimization system demonstrates the potential value of introducing human expertise during automated design. Mount Sinai’s controlled experiment shows the opposite risk: models can absorb irrelevant commercial influence while continuing to produce apparently plausible recommendations.

The metamaterials research offers a physical counterpart to these computational questions. Regularity and uniformity are not always the best solutions. Networks containing structural elements at multiple scales can distribute mechanical loads and damage in ways that repeating geometries may not achieve.

None of these findings should be interpreted as evidence that laboratory demonstrations have already become mature applications. The next steps involve independent validation, improved measurement, engineering scalability and evaluation under realistic operating conditions.

The RKE Signal for November 18, 2026: progress increasingly depends on understanding selective connections, whether they transport proteins into the brain, distribute forces through materials or determine how humans and intelligent systems make decisions together.

RKE | Risolnet Knowledge Engine

Science & Technology News Review | November 18, 2026

risolnet.com