Computers Math

University of Minnesota Twin Cities researchers have constructed a robot that uses machine learning to fully automate a complicated microinjection process used in genetic research.
In their experiments, the researchers were able to use this automated robot to manipulate the genetics of multicellular organisms, including fruit fly and zebrafish embryos. The technology will save labs time and money while enabling them to more easily conduct new, large-scale genetic experiments that were not possible previously using manual techniques
The research is featured on the cover of the April 2024 issue of GENETICS, a peer-reviewed, open access, scientific journal. The work was co-led by two University of Minnesota mechanical engineering graduate students Andrew Alegria and Amey Joshi. The team is also working to commercialize this technology to make it widely available through the University of Minnesota start-up company, Objective Biotechnology.
Microinjection is a method for introducing cells, genetic material, or other agents directly into embryos, cells, or tissues using a very fine pipette. The researchers have trained the robot to detect embryos that are one-hundredth the size of a grain of rice. After detection, the machine can calculate a path and automate the process of the injections.
“This new process is more robust and reproducible than manual injections,” said Suhasa Kodandaramaiah, a University of Minnesota mechanical engineering associate professor and senior author of the study. “With this model, individual laboratories will be able to think of new experiments that you couldn’t do without this type of technology.”
Typically, this type of research requires highly skilled technicians to perform the microinjection, which many laboratories do not have. This new technology could expand the ability to perform large experiments in labs, while reducing time and costs.
“This is very exciting for the world of genetics. Writing and reading DNA have drastically improved in recent years, but having this technology will increase our ability to perform large-scale genetic experiments in a wide range of organisms,” said Daryl Gohl, a co-author of the study, the group leader of the University of Minnesota Genomics Center’s Innovation Lab and research assistant professor in the Department of Genetics, Cell Biology, and Development.

Not only can this technology be used in genetic experiments, but it can also help to preserve endangered species through cryopreservation, a preservation technique conducted at ultra-low temperatures.
“You can use this robot to inject nanoparticles into cells and tissues that helps in cryopreservation and in the process of rewarming afterwards,” Kodandaramaiah explained.
Other team members highlighted other applications for the technology that could have even more impact.
“We hope that this technology could eventually be used for in vitro fertilization, where you could detect those eggs on the microscale level,” said Andrew Alegria, co-lead author on the paper and University of Minnesota mechanical engineering graduate research assistant in the Biosensing and Biorobotics Lab. More

A collaborative research team from NIMS and Tokyo University of Science has successfully developed a cutting-edge artificial intelligence (AI) device that executes brain-like information processing through few-molecule reservoir computing. This innovation utilizes the molecular vibrations of a select number of organic molecules. By applying this device for the blood glucose level prediction in patients with diabetes, it has significantly outperformed existing AI devices in terms of prediction accuracy.
With the expansion of machine learning applications in various industries, there’s an escalating demand for AI devices that are not only highly computational but also feature low-power consumption and miniaturization. Research has shifted towards physical reservoir computing, leveraging physical phenomena presented by materials and devices for neural information processing. One challenge that remains is the relatively large size of the existing materials and devices.
Our research has pioneered the world’s first implementation of physical reservoir computing that operates on the principle of surface-enhanced Raman scattering, harnessing the molecular vibrations of merely a few organic molecules. The information is inputted through ion-gating, which modulates the adsorption of hydrogen ions onto organic molecules (p-mercaptobenzoic acid, pMBA) by applying voltage. The changes in molecular vibrations of the pMBA molecules, which vary with hydrogen ion adsorption, serve the function of memory and nonlinear waveform transformation for calculation. This process, using a sparse assembly of pMBA molecules, has learned approximately 20 hours of a diabetic patient’s blood glucose level changes and managed to predict subsequent fluctuations over the next 5 minutes with an error reduction of about 50% compared to the highest accuracy achieved by similar devices to date.
The outcome of this study indicates that a minimal quantity of organic molecules can effectively perform computations comparable to a computer. This technological breakthrough of conducting sophisticated information processing with minimal materials and in tiny spaces presents substantial practical benefits. It paves the way for the creation of low-power AI terminal devices that can be integrated with a variety of sensors, opening avenues for broad industrial use. More

Creating robots to safely aid disaster victims is one challenge; executing flexible robot control that takes advantage of the material’s softness is another. The use of pliable soft materials to collaborate with humans and work in disaster areas has drawn much recent attention. However, controlling soft dynamics for practical applications has remained a significant challenge.
In collaboration with the University of Tokyo and Bridgestone Corporation, Kyoto University has now developed a method to control pneumatic artificial muscles, which are soft robotic actuators. Rich dynamics of these drive components can be exploited as a computational resource.
“We’ve demonstrated the actuator’s capability to autonomously generate diverse dynamics, including rhythmic patterns and chaos,” explains Nozomi Akashi of KyotoU’s Graduate School of Informatics.
Traditionally, patterns were generated by externally attaching oscillators to robots, enabling locomotion and repetitive motions. However, these oscillators should be removed from the robot to retain their softness. Akashi’s team addresses this difficult issue to bring out the soft robots’ potential.
“In addition, the pattern-changing bifurcation structures can be embedded into the robotic actuator itself,” says Kohei Nakajima of the University of Tokyo’s Graduate School of Information Science and Technology.
The findings suggest that robots can generate qualitatively different patterns outside the learning data, paving the way for the development of robots capable of more adaptable and flexible movements.
“This could streamline the hardware and software development process, making it more efficient and effective,” concludes Akashi. More

Light-driven molecular motors were first developed nearly 25 years ago at the University of Groningen, the Netherlands. This resulted in a shared Nobel Prize for Chemistry for Professor Ben Feringa in 2016. However, making these motors do actual work proved to be a challenge. A new paper from the Feringa lab, published in Nature Chemistry on 26 April, describes a combination of improvements that brings real-life applications closer.
First author Jinyu Sheng, now a postdoctoral researcher at the Institute of Science and Technology Austria (ISTA), adapted a ‘first generation’ light-driven molecular motor during his PhD studies in the Feringa laboratory. His main focus was to increase the efficiency of the motor molecule. ‘It is very fast, but only 2% of the photons that the molecule absorbs drive the rotary movement.’
Increased efficiency
This poor efficiency can get in the way of real-life applications. ‘Besides, increased efficiency would give us better control of the motion,’ adds Sheng. The rotary motion of Feringa’s molecular motor takes place in four steps: two of them are photochemical, while two are temperature-driven. The latter are unidirectional, but the photochemical steps cause an isomerization of the molecule that is usually reversible.
Sheng set out to improve the percentage of absorbed photons that drive rotary motion. ‘It is very difficult to predict how this can be done and, in the end, we accidently discovered a method that worked.’ Sheng added an aldehyde functional group to the motor molecule, as a first step in further transformation. ‘However, I decided to test the motor function of this intermediate version and found it to be very efficient in a way that we had never seen before.’
For this, he cooperated with the Molecular Photonics group at the University of Amsterdam’s Van ‘t Hoff Institute for Molecular Sciences. Using advanced laser spectroscopy and quantum chemical calculations the electronic decay pathways were mapped, providing detailed insight in the working of the molecular motor.
Rotation cycle
Furthermore, it became clear that the adaption indeed gave Sheng better control of the molecule’s rotary movement. As mentioned before, the molecular motor rotates in four discrete steps. Sheng: ‘Previously, if we irradiated a batch of motors with light, we would get a mixture of motors at different stages of the rotation cycle. After the modification, it was possible to synchronize all motors and control them at each stage.’

This opens up all kinds of possibilities. For example, the motors could be used as a chiral dopant in liquid crystals, where the different positions would create different reflection colours. In the Nature Chemistry paper, Sheng and his colleagues present an example of this. Other applications could, for example, be the control of molecular self-assembly.
Applications
The addition of an aldehyde group to the motor molecule also has another interesting effect: it shifts the absorption of light to a longer wavelength. Since longer wavelengths penetrate further into living tissue or bulk material, this means that the motors could work much more efficiently in medical applications and materials science because more light will reach the motor molecule, while this will also use the photons more efficiently.
‘A number of our colleagues are now working with us on this new molecular motor for different applications,’ says Sheng. He expects more papers on this topic in the near future. Meanwhile, there is another challenge for the Feringa lab: ‘The molecular motor is now more efficient but we don’t exactly know why the modification causes this effect. We are currently working on it!’ More

Researchers have developed tiny, flexible devices that can wrap around individual nerve fibres without damaging them.
The researchers, from the University of Cambridge, combined flexible electronics and soft robotics techniques to develop the devices, which could be used for the diagnosis and treatment of a range of disorders, including epilepsy and chronic pain, or the control of prosthetic limbs.
Current tools for interfacing with the peripheral nerves — the 43 pairs of motor and sensory nerves that connect the brain and the spinal cord — are outdated, bulky and carry a high risk of nerve injury. However, the robotic nerve ‘cuffs’ developed by the Cambridge team are sensitive enough to grasp or wrap around delicate nerve fibres without causing any damage.
Tests of the nerve cuffs in rats showed that the devices only require tiny voltages to change shape in a controlled way, forming a self-closing loop around nerves without the need for surgical sutures or glues.
The researchers say the combination of soft electrical actuators with neurotechnology could be an answer to minimally invasive monitoring and treatment for a range of neurological conditions. The results are reported in the journal Nature Materials.
Electric nerve implants can be used to either stimulate or block signals in target nerves. For example, they might help relieve pain by blocking pain signals, or they could be used to restore movement in paralysed limbs by sending electrical signals to the nerves. Nerve monitoring is also standard surgical procedure when operating in areas of the body containing a high concentration of nerve fibres, such as anywhere near the spinal cord.
These implants allow direct access to nerve fibres, but they come with certain risks. “Nerve implants come with a high risk of nerve injury,” said Professor George Malliaras from Cambridge’s Department of Engineering, who led the research. “Nerves are small and highly delicate, so anytime you put something large, like an electrode, in contact with them, it represents a danger to the nerves.”
“Nerve cuffs that wrap around nerves are the least invasive implants currently available, but despite this they are still too bulky, stiff and difficult to implant, requiring significant handling and potential trauma to the nerve,” said co-author Dr Damiano Barone from Cambridge’s Department of Clinical Neurosciences.

The researchers designed a new type of nerve cuff made from conducting polymers, normally used in soft robotics. The ultra-thin cuffs are engineered in two separate layers. Applying tiny amounts of electricity — just a few hundred millivolts — causes the devices to swell or shrink.
The cuffs are small enough that they could be rolled up into a needle and injected near the target nerve. When activated electrically, the cuffs will change their shape to wrap around the nerve, allowing nerve activity to be monitored or altered.
“To ensure the safe use of these devices inside the body, we have managed to reduce the voltage required for actuation to very low values,” said Dr Chaoqun Dong, the paper’s first author. “What’s even more significant is that these cuffs can change shape in both directions and be reprogrammed. This means surgeons can adjust how tightly the device fits around a nerve until they get the best results for recording and stimulating the nerve.”
Tests in rats showed that the cuffs could be successfully placed without surgery, and they formed a self-closing loop around the target nerve. The researchers are planning further testing of the devices in animal models, and are hoping to begin testing in humans within the next few years.
“Using this approach, we can reach nerves that are difficult to reach through open surgery, such as the nerves that control, pain, vision or hearing, but without the need to implant anything inside the brain,” said Barone. “The ability to place these cuffs so they wrap around the nerves makes this a much easier procedure for surgeons, and it’s less risky for patients.”
“The ability to make an implant that can change shape through electrical activation opens up a range of future possibilities for highly targeted treatments,” said Malliaras. “In future, we might be able to have implants that can move through the body, or even into the brain — it makes you dream how we could use technology to benefit patients in future.”
The research was supported in part by the Swiss National Science Foundation, the Cambridge Trust, and the Engineering and Physical Sciences Research Council (EPSRC), part of UK Research and Innovation (UKRI). More

Computer vision can be a valuable tool for anyone tasked with analyzing hours of footage because it can speed up the process of identifying individuals. For example, law enforcement may use it to perform a search for individuals with a simple query, such as “Locate anyone wearing a red scarf over the past 48 hours.”
With video surveillance becoming more and more ubiquitous, Assistant Professor Yogesh Rawat, a researcher at the UCF Center for Research in Computer Vision (CRCV), is working to address privacy issues with advanced software installed on video cameras. His work is supported by $200,000 in funding from the U.S. National Science Foundation’s Accelerating Research Translation (NSF ART) program.
“Automation allows us to watch a lot of footage, which is not possible by humans,” Rawat says. “Surveillance is important for society, but there are always privacy concerns. This development will enable surveillance with privacy preservation.”
His video monitoring software protects the privacy of those recorded by obscuring select elements, such as faces or clothing, both in recordings and in real time. Rawat explains that his software adds perturbations to the RGB pixels in the video feed — the red, green and blue colors of light — so that human eyes are unable to recognize them.
“Mainly we are interested in any identifiable information that we can visually interpret,” Rawat says. “For example, for a person’s face, I can say ‘This is that individual,’ just by identifying the face. It could be the height as well, maybe hair color, hair style, body shape — all those things that can be used to identify any person. All of this is private information.”
Since Rawat aims to have the technology available in edge devices, devices that are not dependent on an outside server such as drones and public surveillance cameras, he and his team are also working on developing the technology so that it’s fast enough to analyze the feed as it is received. This poses the additional challenge of developing algorithms that can process the data as quickly as possible, so that graphics processing units (GPUs) and central processing units (CPUs) can handle the workload of analyzing footage as it is captured.
To that end, his main considerations in implementing the software are speed and size.

“We want to do this very efficiently and very quickly in real time,” Rawat says. “We don’t want to wait for a year, a month or days. We also don’t want to take a lot of computing power. We don’t have a lot of computing power in very small GPUs or very small CPUs. We are not working with large computers there, but very small devices.”
The funding from the NSF ART program will allow Rawat to identify potential users of the technology, including nursing homes, childcare centers and authorities using surveillance cameras. Rawat is one of two UCF researchers to have projects initially funded through the $6 million grant awarded to the university earlier this year. Four more projects will be funded over the next four years.
His work builds on several previous projects spearheaded by other CRCV members, including founder Mubarak Shah and researcher Chen Chen, including extensive work that allows analysis of untrimmed security videos, training artificial intelligence models to operate on a smaller scale and a patent on software that allows for the detection of multiple actions, persons and objects of interest. Funding sources for these works include $3.9 million from the IARPA Biometric Recognition and Identification at Altitude and Range program, $2.8 million from Intelligence Advanced Research Projects Activity (IARPA) Deep Intermodal Video Analysis, and $475,000 from the U.S Combating Terrorism Technical Support Office.
Rawat says his work in computer vision is motivated by a drive to improve our world.
“I’m really interested in understanding how we can easily navigate in this world as humans,” he says. “Visual perception is something I’m very interested in studying, including how we can bring it to machines and make things easy for us as humans and as a society.” More

The IRIS infrared beamline at the BESSY II storage ring now offers a fourth option for characterising materials, cells and even molecules on different length scales. The team has extended the IRIS beamline with an end station for nanospectroscopy and nanoimaging that enables spatial resolutions down to below 30 nanometres. The instrument is also available to external user groups.
The infrared beamline IRIS at the BESSY II storage ring is the only infrared beamline in Germany that is also available to external user groups and is therefore in great demand. Dr Ulrich Schade, in charge of the beamline, and his team continue to develop the instruments to enable unique, state-of-the-art experimental techniques in IR spectroscopy.
As part of a recent major upgrade to the beamline, the team, together with the Institute of Chemistry at Humboldt University Berlin, has built an additional infrared near-field microscope.
“With the nanoscope, we can resolve structures smaller than a thousandth of the diameter of a human hair and thus reach the innermost structures of biological systems, catalysts, polymers and quantum materials,” says Dr Alexander Veber, who led this extension.
The new nanospectroscopy end station is based on a scanning optical microscope and enables imaging and spectroscopy with infrared light with a spatial resolution of more than 30 nm. To demonstrate the performance of the new end station, Veber analysed individual cellulose microfibrils and imaged cell structures. All end stations are available to national and international user groups. More

Machines can learn not only to make predictions, but also to handle causal relationships. An international research team shows how this could make therapies safer, more efficient, and more individualized.
Artificial intelligence is making progress in the medical arena. When it comes to imaging techniques and the calculation of health risks, there is a plethora of AI methods in development and testing phases. Wherever it is a matter of recognizing patterns in large data volumes, it is expected that machines will bring great benefit to humanity. Following the classical model, the AI compares information against learned examples, draws conclusions, and makes extrapolations.
Now an international team led by Professor Stefan Feuerriegel, Head of the Institute of Artificial Intelligence (AI) in Management at LMU, is exploring the potential of a comparatively new branch of AI for diagnostics and therapy. Can causal machine learning (ML) estimate treatment outcomes — and do so better than the ML methods generally used to date? Yes, says a landmark study by the group, which has been published in the journal Nature Medicine: causal ML can improve the effectiveness and safety of treatments.
In particular, the new machine learning variant offers “an abundance of opportunities for personalizing treatment strategies and thus individually improving the health of patients,” write the researchers, who hail from Munich, Cambridge (United Kingdom), and Boston (United States) and include Stefan Bauer and Niki Kilbertus, professors of computer science at the Technical University of Munich (TUM) and group leaders at Helmholtz AI.
As regards machine assistance in therapy decisions, the authors anticipate a decisive leap forward in quality. Classical ML recognizes patterns and discovers correlations, they argue. However, the causal principle of cause and effect remains closed to machines as a rule; they cannot address the question of why. And yet many questions that arise when making therapy decisions contain causal problems within them. The authors illustrate this with the example of diabetes: Classical ML would aim to predict how probable a disease is for a given patient with a range of risk factors. With causal ML, it would ideally be possible to answer how the risk changes if the patient gets an anti-diabetes drug; that is, gauge the effect of a cause (prescription of medication). It would also be possible to estimate whether another treatment plan would be better, for example, than the commonly prescribed medication, metformin.
To be able to estimate the effect of a — hypothetical — treatment, however, “the AI models must learn to answer questions of a ‘What if?’ nature,” says Jonas Schweisthal, doctoral candidate in Feuerriegel’s team. “We give the machine rules for recognizing the causal structure and correctly formalizing the problem,” says Feuerriegel. Then the machine has to learn to recognize the effects of interventions and understand, so to speak, how real-life consequences are mirrored in the data that has been fed into the computers.
Even in situations for which reliable treatment standards do not yet exist or where randomized studies are not possible for ethical reasons because they always contain a placebo group, machines could still gauge potential treatment outcomes from the available patient data and thus form hypotheses for possible treatment plans, so the researchers hope. With such real-world data, it should generally be possible to describe the patient cohorts with ever greater precision in the estimates, thereby bringing individualized therapy decisions that much closer. Naturally, there would still be the challenge of ensuring the reliability and robustness of the methods.
“The software we need for causal ML methods in medicine doesn’t exist out of the box,” says Feuerriegel. Rather, “complex modeling” of the respective problem is required, involving “close collaboration between AI experts and doctors.” Like his TUM colleagues Stefan Bauer and Niki Kilbertus, Feuerriegel also researches questions relating to AI in medicine, decision-making, and other topics at the Munich Center for Machine Learning (MCML) and the Konrad Zuse School of Excellence in Reliable AI. In other fields of application, such as marketing, explains Feuerriegel, the work with causal ML has already been in the testing phase for some years now. “Our goal is to bring the methods a step closer to practice. The paper describes the direction in which things could move over the coming years.” More