Computers Math

A new computer process developed by chemists at ETH Zurich makes it possible to generate active pharmaceutical ingredients quickly and easily based on a protein’s three-​dimensional surface. The new process could revolutionise drug research.
“It’s a real breakthrough for drug discovery,” says Gisbert Schneider, Professor at ETH Zurich’s Department of Chemistry and Applied Biosciences. Together with his former doctoral student Kenneth Atz, he has developed an algorithm that uses artificial intelligence (AI) to design new active pharmaceutical ingredients. For any protein with a known three-dimensional shape, the algorithm generates the blueprints for potential drug molecules that increase or inhibit the activity of the protein. Chemists can then synthesise and test these molecules in the laboratory.
All the algorithm needs is a protein’s three-dimensional surface structure. Based on that, it designs molecules that bind specifically to the protein according to the lock-and-key principle so they can interact with it.
Excluding side effects from the outset
The new method builds on the decades-long efforts of chemists to elucidate the three-dimensional structure of proteins and to use computers to search for suitable potential drug molecules. Until now, this has often involved laborious manual work, and in many cases the search yielded molecules that were very difficult or impossible to synthesise. If researchers used AI in this process at all in recent years, it was primarily to improve existing molecules.
Now, without human intervention, a generative AI is able to develop drug molecules from scratch that match a protein structure. This groundbreaking new process ensures right from the start that the molecules can be chemically synthesised. In addition, the algorithm suggests only molecules that interact with the specified protein at the desired location and hardly at all with any other proteins. “This means that when designing a drug molecule, we can be sure that it has as few side effects as possible,” Atz says.
To create the algorithm, the scientists trained an AI model with information from hundreds of thousands of known interactions between chemical molecules and the corresponding three-dimensional protein structures.

Successful tests with industry
Together with researchers from the pharmaceutical company Roche and other cooperation partners, the ETH team tested the new process and demonstrated what it is capable of. The scientists searched for molecules that interact with proteins in the PPAR class — proteins that regulate sugar and fatty acid metabolism in the body. Several diabetes drugs used today increase the activity of PPARs, which causes the cells to absorb more sugar from the blood and the blood sugar level to fall.
Straightaway the AI designed new molecules that also increase the activity of PPARs, like the drugs currently available, but without a lengthy discovery process. After the ETH researchers had produced these molecules in the lab, colleagues at Roche subjected them to a variety of tests. These showed that the new substances are indeed stable and non-toxic right from the start.
The researchers aren’t now pursuing these molecules any further with a view to bringing drugs based on them to the market. Instead, the purpose of the molecules was to subject the new AI process to an initial rigorous test. Schneider says, however, that the algorithm is already being used for similar studies at ETH Zurich and in industry. One of these is a project with the Children’s Hospital Zurich for the treatment of medulloblastomas, the most common malignant brain tumours in children. Moreover, the researchers have published the algorithm and its software so that researchers worldwide can now use them for their own projects.
“Our work has made the world of proteins accessible for generative AI in drug research,” Schneider says. “The new algorithm has enormous potential.” This is especially true for all medically relevant proteins in the human body that don’t interact with any known chemical compounds. More

An ongoing research project at Tampere University aims to create adaptable safety systems for highly automated off-road mobile machinery to meet industry needs. Research has revealed critical gaps in compliance with legislation related to public safety when using mobile working machines controlled by artificial intelligence.
As the adoption of highly automated off-road machinery increases, so does the need for robust safety measures. Conventional safety processes often fail to consider the health and safety risks posed by systems controlled by artificial intelligence (AI).
Marea de Koning, a doctoral researcher specialising in automation at Tampere University, conducts research with the aim of ensuring public safety without compromising technological advancements by developing a safety framework specifically tailored for autonomous mobile machines operating in collaboration with humans. This framework intents to enable original equipment manufacturers (OEM), safety & system engineers, and industry stakeholders to create safety systems that comply with evolving legislation.
Balance between humans and autonomous machines
Anticipating all the possible ways a hazard can emerge and ensuring that the AI can safely manage hazardous scenarios is practically impossible. We need to adjust our approach to safety to focus more on finding ways to successfully manage unforeseen events.
We need robust risk management systems, often incorporating a human-in-the-loop safety option. Here a human supervisor is expected to intervene when necessary. But in autonomous machinery, relying on human intervention is impractical. According to de Koning, there can be measurable degradations in human performance when automation is used due to, for example, boredom, confusion, cognitive capacities, loss of situational awareness, and automation bias. These factors significantly impact safety, and a machine must become capable of safely managing its own behaviour.
“Myapproach considers hazards with AI-driven decision-making, risk assessment, and adaptability to unforeseen scenarios. I think it is important to actively engage with industry partners to ensure real-world applicability. By collaborating with manufacturers, it is possible to bridge the gap between theoretical frameworks and practical implementation,” she says.

The framework intents to support OEMs in designing and developing compliant safety systems and ensure that their products adhere to evolving regulations.
Integrating framework to existing machinery
Marea de Koning started her research in November 2020 and will finish it by November 2024. The project is funded partly by the Doctoral School of Industry Innovations and partly by a Finnish system supplier.
De Koning’s next research project, starting in April, will focus on integrating a subset of her safety framework and rigorously testing its effectiveness. Regulation 2023/1230 replaces Directive 2006/42/ec as of January 2027, significantly challenging OEMs.
“I’m doing everything I can to ensure that safety remains at the forefront of technological advancements,” she concludes.
The research provides valuable insights for policymakers, engineers and safety professionals. The article presenting the findings titled A Comprehensive Approach to Safety for Highly Automated Off-Road Machinery under Regulation 2023/1230 was published in the prestigious Journal of Safety Science. More

Researchers from Aston University have found that people following healthy eating accounts on social media for as little as two weeks ate more fruit and vegetables and less junk food.
Previous research has shown that positive social norms about fruit and vegetables increases individuals’ consumption. The research team sought to investigate whether positive representation of healthier food on social media would have the same effect. The research was led by Dr Lily Hawkins, whose PhD study it was, supervised by Dr Jason Thomas and Professor Claire Farrow in the School of Psychology.
The researchers recruited 52 volunteers, all social media users, with a mean age of 22, and split them into two groups. Volunteers in the first group, known as the intervention group, were asked to follow healthy eating Instagram accounts in addition to their usual accounts. Volunteers in the second group, known as the control group, were asked to follow interior design accounts. The experiment lasted two weeks, and the volunteers recorded what they ate and drank during the time period.
Overall, participants following the healthy eating accounts ate an extra 1.4 portions of fruit and vegetables per day and 0.8 fewer energy dense items, such as high-calorie snacks and sugar-sweetened drinks, per day. This is a substantial improvement compared to previous educational and social media-based interventions attempting to improve diets.
Dr Thomas and the team believe affiliation is a key component of the change in eating behaviour. For example, the effect was more pronounced amongst participants who felt affiliated with other Instagram users.
The 2018 NHS Health Survey for England study showed that only 28% of the UK population consumed the recommended five portions of fruit and vegetables per day. Low consumption of such food is linked to heart disease, cancer and stroke, so identifying ways to encourage higher consumption is vital. Exposing people to positive social norms, using posters in canteens encouraging vegetable consumption, or in bars to discourage dangerous levels of drinking, have been shown to work. Social media is so prevalent now that the researchers believe it could be an ideal way to spread positive social norms around high fruit and vegetable consumption, particularly amongst younger people.
Dr Thomas said:
“This is only a pilot intervention study at the moment, but it’s quite an exciting suite of findings, as it suggests that even some minor tweaks to our social media accounts might lead to substantial improvements in diet, at zero cost! Our future work will examine whether such interventions actually do change our perceptions of what others are consuming, and also, whether these interventions produce effects that are sustained over time.”
Dr Hawkins, who is now at the University of Exeter, said:
“Our previous research has demonstrated that social norms on social media may nudge food consumption, but this pilot demonstrates that this translates to the real world. Of course, we would like to now understand whether this can be replicated in a larger, community sample.” More

A computer game helped upper secondary school students become better at distinguishing between reliable and misleading news. This is shown by a study conducted by researchers at Uppsala University and elsewhere.
“This is an important step towards equipping young people with the tools they need to navigate in a world full of disinformation. We all need to become better at identifying manipulative strategies — prebunking, as it is known — since it is virtually impossible to discern deep fakes, for example, and other AI-generated disinformation with the naked eye,” says Thomas Nygren, Professor of Education at Uppsala University.
Along with three other researchers, he conducted a study involving 516 Swedish upper secondary school students in different programmes at four schools. The study, published in the Journal of Research on Technology in Education, investigated the effect of the game Bad News in a classroom setting — this is the first time the game has been scientifically tested in a normal classroom. The game has been created for research and teaching, and the participants assume the role of spreader of misleading news. The students in the study either played the game individually, in pairs or in whole class groups with a shared scorecard — all three methods had positive effects. This surprised the researchers, who believed students would learn more by working at the computer together.
“The students improved their ability to identify manipulative techniques in social media posts and to distinguish between reliable and misleading news,” Nygren comments.
The study also showed that students who already had a positive attitude towards trustworthy news sources were better at distinguishing disinformation, and this attitude became significantly more positive after playing the game. Moreover, many students improved their assessments of credibility and were able to explain how they could identify manipulative techniques in a more sophisticated way.
The researchers noted that competitive elements in the game made for greater interest and enhanced its benefit. They therefore conclude that the study contributes insights for teachers into how serious games can be used in formal instruction to promote media and information literacy.
“Some people believe that gamification can enhance learning in school. However, our results show that more gamification in the form of competitive elements does not necessarily mean that students learn more — though it can be perceived as more fun and interesting,” Nygren says.
Participating researchers: Carl-Anton Werner Axelsson (Mälardalen and Uppsala), Thomas Nygren (Uppsala), Jon Roozenbeek (Cambridge) and Sander van der Linden (Cambridge). More

Setting the stage for a new era of immersive displays, researchers are one step closer to mixing the real and virtual worlds in an ordinary pair of eyeglasses using high-definition 3D holographic images, according to a study led by Princeton University researchers.
Holographic images have real depth because they are three dimensional, whereas monitors merely simulate depth on a 2D screen. Because we see in three dimensions, holographic images could be integrated seamlessly into our normal view of the everyday world.
The result is a virtual and augmented reality display that has the potential to be truly immersive, the kind where you can move your head normally and never lose the holographic images from view. “To get a similar experience using a monitor, you would need to sit right in front of a cinema screen,” said Felix Heide, assistant professor of computer science and senior author on a paper published April 22 in Nature Communications.
And you wouldn’t need to wear a screen in front of your eyes to get this immersive experience. Optical elements required to create these images are tiny and could potentially fit on a regular pair of glasses. Virtual reality displays that use a monitor, as current displays do, require a full headset. And they tend to be bulky because they need to accommodate a screen and the hardware necessary to operate it.
“Holography could make virtual and augmented reality displays easily usable, wearable and ultrathin,” said Heide. They could transform how we interact with our environments, everything from getting directions while driving, to monitoring a patient during surgery, to accessing plumbing instructions while doing a home repair.
One of the most important challenges is quality. Holographic images are created by a small chip-like device called a spatial light modulator. Until now, these modulators could only create images that are either small and clear or large and fuzzy. This tradeoff between image size and clarity results in a narrow field of view, too narrow to give the user an immersive experience. “If you look towards the corners of the display, the whole image may disappear,” said Nathan Matsuda, research scientist at Meta and co-author on the paper.
Heide, Matsuda and Ethan Tseng, doctoral student in computer science, have created a device to improve image quality and potentially solve this problem. Along with their collaborators, they built a second optical element to work in tandem with the spatial light modulator. Their device filters the light from the spatial light modulator to expand the field of view while preserving the stability and fidelity of the image. It creates a larger image with only a minimal drop in quality.
Image quality has been a core challenge preventing the practical applications of holographic displays, said Matsuda. “The research brings us one step closer to resolving this challenge,” he said.
The new optical element is like a very small custom-built piece of frosted glass, said Heide. The pattern etched into the frosted glass is the key. Designed using AI and optical techniques, the etched surface scatters light created by the spatial light modulator in a very precise way, pushing some elements of an image into frequency bands that are not easily perceived by the human eye. This improves the quality of the holographic image and expands the field of view.
Still, hurdles to making a working holographic display remain. The image quality isn’t yet perfect, said Heide, and the fabrication process for the optical elements needs to be improved. “A lot of technology has to come together to make this feasible,” said Heide. “But this research shows a path forward.” More

Researchers at the LMU Equine Clinic have developed a deep learning tool that is capable of reliably diagnosing moon blindness in horses based on photos.
Colloquially known as moon blindness, equine recurrent uveitis (ERU) is an inflammatory ocular disease in horses, which can lead to blindness or loss of the affected eye. It is one of the most common eye diseases in horses and has a major economic impact. Correct and swift diagnosis is very important to minimize lasting damage. A team led by Professor Anna May from the LMU Equine Clinic has developed and trained a deep learning tool that reliably recognizes the disease and can support veterinary doctors in the making of diagnoses, as the researchers report in a current study.
In an online survey, the researchers asked some 150 veterinarians to evaluate 40 photos. The pictures showed a mixture of healthy eyes, eyes with ERU, and eyes with other diseases. Working on the basis of image analyses, the deep learning tool was given the task of evaluating the same photos. Subsequently, May compared the results of the veterinarians against those of the AI. She discovered that veterinary doctors specialized in horses interpreted the pictures correctly 76 percent of the time, while the remaining vets from small animal or mixed practices were right 67 percent of the time. “With the deep learning tool, the probability of getting a correct answer was 93 percent,” says May. “Although the differences were not statistically significant, they nonetheless show that the AI reliably recognizes an ERU and has great potential as a tool for supporting veterinary doctors.”
The tool is web-app-based and simple to use. All you need is a smartphone. “It’s not meant to replace veterinarians, but can help them reach the correct diagnosis. It is particularly valuable for less experienced professionals or for horse owners in regions where vets are few and far between,” emphasizes May. Through the early detection of ERU, affected horses can receive appropriate treatment more quickly, which can be decisive in slowing down the progress of the disease and saving the afflicted eyes. More

Magnetic cilia — artificial hairs whose movement is powered by embedded magnetic particles — have been around for a while, and are of interest for applications in soft robotics, transporting objects and mixing liquids. However, existing magnetic cilia move in a fixed way. Researchers have now demonstrated a technique for creating magnetic cilia that can be “reprogrammed,” changing their magnetic properties at room temperature to change the motion of the cilia as needed.
Most magnetic cilia make use of ‘soft’ magnets, which do not generate a magnetic field but become magnetic in the presence of a magnetic field. Only a few previous magnetic cilia have made use of ‘hard’ magnets, which are capable of producing their own magnetic field. One of the advantages of using hard magnets is that they can be programmed, meaning that you can give the magnetic field generated by the material a specific polarization. Controlling the magnetic polarization — or magnetization — allows you to essentially dictate precisely how the cilia will flex when an external magnetic field is applied.
“What’s novel about this work, is that we have demonstrated a technique that allows us to not only program magnetic cilia, but also controllably reprogram them,” says Joe Tracy, corresponding author of a paper on the work and professor of materials science and engineering at North Carolina State University. “We can change the direction of the material’s magnetization at room temperature, which in turn allows us to completely change how the cilia flex. It’s like getting a swimmer to change their stroke.”
For this work, the researchers created magnetic cilia consisting of a polymer embedded with magnetic microparticles. Specifically, the microparticles are neodymium magnets — powerful magnets made of neodymium, iron and boron.
To make the cilia, the researchers introduce the magnetic microparticles into a polymer dissolved in a liquid. This slurry is then exposed to an electromagnetic field that is sufficiently powerful to give all of the microparticles the same magnetization. By then applying a less powerful magnetic field as the liquid polymer dries, the researchers are able to control the behavior of the microparticles, resulting in the formation of cilia that are regularly spaced across the substrate.
“This regularly ordered cilia carpet is initially programmed to behave in a uniform way when exposed to an external magnetic field,” Tracy says. “But what’s really interesting here, is that we can reprogram that behavior, so that the cilia can be repurposed to have a completely different actuation.”
To do that, the researchers first embed the cilia in ice, which fixes all of the cilia in the desired direction. The researchers then expose the cilia to a damped, alternating magnetic field which has the effect of disordering the magnetization of the microparticles. In other words, they substantially erase the preprogrammed magnetization that was shared by all of the microparticles when the cilia were fabricated.

“The reprogramming step is fairly straightforward,” Tracy says. “We apply an oscillating field to reset the magnetization, then apply a strong magnetic field to the cilia which allows us to magnetize the microparticles in a new direction.”
“By mostly erasing the initial magnetization, we’re better able to reprogram the magnetization of the microparticles,” says Matt Clary, first author of the paper and a Ph.D. student at NC State. “We show in this work that if you leave out that erasing step you have less control over the orientation of the microparticles’ magnetization when reprogramming.”
“We also found that when the magnetization of the microparticles is perpendicular to the long axis of the cilia, we can cause the cilia to ‘snap’ in a rotating field, meaning they abruptly change their orientation,” says Tracy.
In addition, the research team developed a computational model that allows users to predict the bending behavior of magnetic cilia based on hard magnets, depending on the orientation of the cilia’s polarization.
“This model could be used in the future to guide the design of hard-magnetic cilia and related soft actuators,” says Ben Evans, coauthor of the paper and professor of physics at Elon University.
“Ultimately, we think this work is valuable to the field because it allows repurposing of magnetic cilia for new functions or applications, especially in remote environments,” Tracy says. “Methods developed in this work may also be applied to the broader field of magnetic soft actuators.”
The paper, “Magnetic Reprogramming of Self-Assembled Hard-Magnetic Cilia,” is published open access in the journal Advanced Materials Technologies. The paper was co-authored by Saarah Cantu, a former graduate student at NC State; and Jessica Liu, a former Ph.D. student at NC State.
This work was done with support from the National Science Foundation, under grants 1663416 and 1662641; and from the Higher Education Emergency Relief Fund. More

The mass production of conventional silicon chips relies on a successful business model with large ‘semiconductor fabrication plants’ or ‘foundries’. New research by KU Leuven and imec shows that this ‘foundry’ model can also be applied to the field of flexible, thin-film electronics. Adopting this approach would give innovation in the field a huge boost.
Silicon semiconductors have become the ‘oil’ of the computer age, which was also demonstrated recently by the chip shortage crisis. However, one of the disadvantages of conventional silicon chips is that they’re not mechanically flexible. On the other hand you have the field of flexible electronics, which is driven by an alternative semiconductor technology: the thin-film transistor, or TFT. The applications in which TFTs can be used are legion: from wearable healthcare patches and neuroprobes over digital microfluidics and robotic interfaces to bendable displays and Internet of Things (IoT) electronics.
TFT technology has well evolved, but unlike with conventional semiconductor technology the potential to use it in various applications has barely been exploited. In fact, TFTs are currently mainly mass-produced with the purpose of integrating them in displays of smartphones, laptops and smart TVs — where they are used to control pixels individually. This limits the freedom of chip designers who dream of using TFTs in flexible microchips and to come up with innovative, TFT-based applications. “This field can benefit hugely from a foundry business model similar to that of the conventional chip industry,” says Kris Myny, professor at the KU Leuven’s Emerging technologies, Systems and Security unit in Diepenbeek, and also a guest professor at imec.
Foundry business model
At the heart of the worldwide microchip market is the so-called foundry model. In this business model, large ‘semiconductor fabrication plants’ or ‘foundries’ (like TSMC from Taiwan) focus on the mass production of chips on silicon wafers. These are then used by the foundries’ clients — the companies that design and order the chips — to integrate them in specific applications. Thanks to this business model, the latter companies have access to complex semiconductor manufacturing to design the chips they need.
Myny’s group has now shown that such a business model is also viable in the field of thin-film electronics. They designed a specific TFT-based microprocessor and let it be produced in two foundries, after which they tested it in their lab, with success. The same chip was produced in two versions, based on two separate TFT technologies (using different substrates) that are both mainstream. Their research paper is published in Nature.
Multi-project approach
The microprocessor Myny and his colleagues built is the iconic MOS 6502. Today this chip is a ‘museum piece’, but in the 70s it was the driver of the first Apple, Commodore and Nintendo computers. The group developed the 6502 chip on a wafer (using amorphous indium-gallium-zinc-oxide) and on a plate (using low-temperature polycrystalline silicon). In both cases the chips were manufactured on the substrate together with other chips, or ‘projects’. This ‘multi-project’ approach enables foundries to produce different chips on-demand from designers on single substrates.
The chip Myny’s group made is less than 30 micrometer thick, less than a human hair. That makes it ideal for, for example, medical applications like wearable patches. Such ultra-thin wearables can be used to make electrocardiograms or electromyograms, to study the condition of respectively the heart and muscles. They would feel just like a sticker, while patches with a silicon-based chip always feel knobbly.
Although the performance of the 6502 microprocessor is not comparable with modern ones, this research demonstrates that also flexible chips can be designed and produced in a multi-project approach, analogue to the way this happens in the conventional chip industry. Myny concludes: “We will not compete with silicon-based chips, we want to stimulate and accelerate innovation based on flexible, thin-film electronics.” More