Not that experts haven’t tried.

Sleep analysis, also known as polysomnography, is used to diagnose sleep disorders by recording multiple types of data, including brain (electroencephalogram or EEG) and heart (electrocardiogram or ECG). Typically, patients are hooked up to dozens of sensors and wires in a clinic, tracking brain, eye, muscle, breathing, and heart activity while sleeping. Not exactly Zzz-inducing.

But what if you could perform the same test at home, just as accurately and in real time?

For the first time, computer science researchers at the University of Southern California have developed an approach that matches the performance of expert-scored polysomnography using just a single-lead echocardiogram. The software, which is open-source, allows anyone with basic coding experience to create their own low-cost, DIY sleep-tracking device.

“Researchers have been trying for decades to find simpler and cheaper methods to monitor sleep‚ especially without the awkward cap,” said lead author Adam Jones, who recently earned his PhD from USC. “But so far, the poor performance, even in ideal conditions, has led to the conclusion that it won’t be possible and that measuring brain activity is necessary. Our research shows that this assumption is no longer true.”

The model, which assesses sleep stages at the highest level, also significantly outperformed other EEG-less models, said the researchers, including commercial sleep-tracking devices. “We wanted to develop a system that addresses the limitations of current methods and the need for more accessibility and affordability in sleep analysis,” said Jones.

The study, published June 2024 in the journal Computers in Biology and Medicine, was co-authored by Laurent Itti, a professor of computer science and Jones’ advisor, and Jones’ longtime collaborator, Bhavin R. Sheth, a USC alumnus and electrical engineer at the University of Houston.

Could the heart be leading the band?

Sleep, a key cognitive decline predictor, becomes shorter and more fragmented with age — a finding validated by both previous studies and the researchers’ neural network. But this decline happens earlier than you might expect. A recent study in Neurology found that people who have more interrupted sleep in their 30s and 40s are more than twice as likely to have memory problems a decade later.

Chronic poor sleep can also contribute to the accumulation of beta-amyloid plaques, a hallmark of Alzheimer’s disease.

“It’s a little scary,” said Jones, who admits he was formerly in the “sleep when I’m dead” camp before embarking on this research as a hobby project in 2010. “That’s why I want these interventions to come quickly and to make them accessible to as many people as possible. This software could help tease apart what’s happening when we sleep every night.”

The researchers trained their model on a large, diverse dataset of 4,000 recordings from subjects ranging from 5 to 90 years old, using only heart data and a deep-learning neural network. Through trial and error, spanning hundreds of iterations, they found that the automated ECG-only network could score sleep just as well as the “gold standard” polysomnography. It successfully categorized sleep into all five stages, including rapid eye movement (REM), which is essential for memory consolidation and emotional stability, and non-REM sleep, including deep sleep, which is crucial for physical and mental restoration.

In addition to simplifying a typically expensive and cumbersome process, this insight highlights a deeper connection between the heart and the brain than previously understood. It also underscores the role of the autonomic nervous system, which links the brain and heart.

“The heart and the brain are connected in ways that are not well-understood, and this research aims to bridge that gap,” said Jones. “There is a lot of evidence in my paper that, in fact, the heart may be leading the band, as it were.”

The work could also help improve sleep studies in remote populations, helping to shed light on the origins and functions of sleep.

In a follow-up paper currently being prepared, Jones aims to explore further what the network focuses on in the ECG data. “I think there is a lot of information hidden in the heart that we don’t know about yet,” he said.

Lithium-metal batteries have the potential to achieve energy densities per unit mass higher than those of the lithium-ion batteries currently in use. For this reason, expectations are high for their use in a wide range of technologies, including drones, electric vehicles and household electricity storage systems. In 2018, NIMS and SoftBank established the NIMS-SoftBank Advanced Technologies Development Center. Together they have since carried out research on high-energy-density rechargeable batteries for use in various systems, such as mobile phone base stations, the Internet of Things (IoT) and high altitude platform stations (HAPS).

A lithium-metal battery with an energy density higher than 300 Wh/kg and a life of more than 200 charge/discharge cycles has previously been reported. Putting high-performance lithium-metal batteries like this into practical use while ensuring their safety will require the development of techniques capable of accurately estimating the cycle lives of these batteries. However, degradation mechanisms are more complex in lithium-metal batteries than in conventional lithium-ion batteries and are not yet fully understood, making the development of models capable of predicting the cycle lives of lithium-metal batteries a great challenge.

This research team fabricated a large number of high-energy-density lithium-metal battery cells — each composed of a lithium-metal anode and a nickel-rich cathode — using advanced battery fabrication techniques the team had previously developed. The team then evaluated the charge/discharge performance of these cells. Finally, the team constructed a model able to predict the cycle lives of lithium-metal batteries by applying machine learning methods to the charge/discharge data. The model proved able to make accurate predictions by analyzing charge, discharge and voltage relaxation process data without relying on any assumption about specific battery degradation mechanisms.

The team intends to further improve the cycle life prediction accuracy of the model and expedite efforts to put high-energy-density lithium-metal batteries into practical use by leveraging the model in the development of new lithium-metal anode materials.

Finding the next groundbreaking polymer is always a challenge, but now Georgia Tech researchers are using artificial intelligence (AI) to shape and transform the future of the field. Rampi Ramprasad’s group develops and adapts AI algorithms to accelerate materials discovery.

This summer, two papers published in the Nature family of journals highlight the significant advancements and success stories emerging from years of AI-driven polymer informatics research. The first, featured in Nature Reviews Materials, showcases recent breakthroughs in polymer design across critical and contemporary application domains: energy storage, filtration technologies, and recyclable plastics. The second, published in Nature Communications, focuses on the use of AI algorithms to discover a subclass of polymers for electrostatic energy storage, with the designed materials undergoing successful laboratory synthesis and testing.

“In the early days of AI in materials science, propelled by the White House’s Materials Genome Initiative over a decade ago, research in this field was largely curiosity-driven,” said Ramprasad, a professor in the School of Materials Science and Engineering. “Only in recent years have we begun to see tangible, real-world success stories in AI-driven accelerated polymer discovery. These successes are now inspiring significant transformations in the industrial materials R&D landscape. That’s what makes this review so significant and timely.”

AI Opportunities

Ramprasad’s team has developed groundbreaking algorithms that can instantly predict polymer properties and formulations before they are physically created. The process begins by defining application-specific target property or performance criteria. Machine learning (ML) models train on existing material-property data to predict these desired outcomes. Additionally, the team can generate new polymers, whose properties are forecasted with ML models. The top candidates that meet the target property criteria are then selected for real-world validation through laboratory synthesis and testing. The results from these new experiments are integrated with the original data, further refining the predictive models in a continuous, iterative process.

While AI can accelerate the discovery of new polymers, it also presents unique challenges. The accuracy of AI predictions depends on the availability of rich, diverse, extensive initial data sets, making quality data paramount. Additionally, designing algorithms capable of generating chemically realistic and synthesizable polymers is a complex task.

The real challenge begins after the algorithms make their predictions: proving that the designed materials can be made in the lab and function as expected and then demonstrating their scalability beyond the lab for real-world use. Ramprasad’s group designs these materials, while their fabrication, processing, and testing are carried out by collaborators at various institutions, including Georgia Tech. Professor Ryan Lively from the School of Chemical and Biomolecular Engineering frequently collaborates with Ramprasad’s group and is a co-author of the paper published in Nature Reviews Materials.

“In our day-to-day research, we extensively use the machine learning models Rampi’s team has developed,” Lively said. “These tools accelerate our work and allow us to rapidly explore new ideas. This embodies the promise of ML and AI because we can make model-guided decisions before we commit time and resources to explore the concepts in the laboratory.”

Using AI, Ramprasad’s team and their collaborators have made significant advancements in diverse fields, including energy storage, filtration technologies, additive manufacturing, and recyclable materials.

Polymer Progress

One notable success, described in the Nature Communications paper, involves the design of new polymers for capacitors, which store electrostatic energy. These devices are vital components in electric and hybrid vehicles, among other applications. Ramprasad’s group worked with researchers from the University of Connecticut.

Current capacitor polymers offer either high energy density or thermal stability, but not both. By leveraging AI tools, the researchers determined that insulating materials made from polynorbornene and polyimide polymers can simultaneously achieve high energy density and high thermal stability. The polymers can be further enhanced to function in demanding environments, such as aerospace applications, while maintaining environmental sustainability.

“The new class of polymers with high energy density and high thermal stability is one of the most concrete examples of how AI can guide materials discovery,” said Ramprasad. “It is also the result of years of multidisciplinary collaborative work with Greg Sotzing and Yang Cao at the University of Connecticut and sustained sponsorship by the Office of Naval Research.”

Industry Potential

The potential for real-world translation of AI-assisted materials development is underscored by industry participation in the Nature Reviews Materials article. Co-authors of this paper also include scientists from Toyota Research Institute and General Electric. To further accelerate the adoption of AI-driven materials development in industry, Ramprasad co-founded Matmerize Inc., a software startup company recently spun out of Georgia Tech. Their cloud-based polymer informatics software is already being used by companies across various sectors, including energy, electronics, consumer products, chemical processing, and sustainable materials.

“Matmerize has transformed our research into a robust, versatile, and industry-ready solution, enabling users to design materials virtually with enhanced efficiency and reduced cost,” Ramprasad said. “What began as a curiosity has gained significant momentum, and we are entering an exciting new era of materials by design.”

There is growing consumer demand for honey, with £89.8 million worth of honey imported to the UK in 2023. But as a high-value product it is vulnerable to fraud, with syrups added to dilute the pure honey — a report from the European Commission in 2023 found 46% of 147 honey samples tested were likely to have been adulterated with cheap plant syrups.

Because honey’s characteristics vary due to sources of nectar, season of harvest and geography, it can be very difficult and complex to detect adulterated products. Authentication methods are costly and time consuming, and there is a growing appetite for reliable testing and the adoption of new rules to combat fraud.

Now scientists at Cranfield University have successfully tested two new methods to authenticate UK honey quickly and accurately.

Detecting fake honey without opening the jar

A research project led by Dr Maria Anastasiadi, Lecturer in Bioinformatics at Cranfield University, with the Food Standards Agency and the UK’s Science and Technology Facilities Council (STFC), used a specialist light analysis technique to detect fake honey without opening the jar.

Samples of UK honeys spiked with rice and sugar beet syrups were tested using the non-invasive Spatial Offset Raman Spectroscopy (SORS) method — developed originally at STFC’s Central Laser Facility (CLF) — more commonly used in pharmaceutical and security diagnostics. This proved highly accurate in detecting sugar syrups present in the honey. SORS rapidly identified the ‘fingerprint’ of each ingredient in the product, and the scientists combined this technique with machine learning to successfully detect and identify sugar syrups from various plant sources.

The analysis method is portable and easy to implement, making it an ideal screening tool for testing honey along the supply chain.

Dr Anastasiadi commented: “Honey is expensive, and in demand — and can be targeted by fraudsters which leaves genuine suppliers out of pocket and undermines consumers’ trust. This method is an effective, quick tool to identify suspicious samples of honey, helping the industry to protect consumers and verify supply chains.”

The paper Application of Spatial Offset Raman Spectroscopy (SORS) and Machine Learning for Sugar Syrup Adulteration Detection in UK Honey was published in Foods 2024, vol. 13.

DNA traces in honey used to decipher real from fake

DNA barcoding was used in a second study, in collaboration with the Food Standards Agency and the Institute for Global Food Security at Queen’s University of Belfast, to detect rice and corn syrups spiked in UK honey samples.

Scientists used 17 honey samples collected from bee farmers around the UK, representing different seasons and floral nectar sources, and bought four samples of UK honey from supermarkets and online retailers. The samples were then spiked with corn and rice syrups produced in a range of countries.

DNA barcoding — a method already used in food authentication to identify plant species in products — was effective in breaking down the composition of each sample, to successfully detect syrups even at 1% adulteration level.

“To date, DNA methods haven’t been widely used to examine honey authenticity,” commented Dr Anastasiadi. “But our study showed that this is a sensitive, reliable and robust way to detect adulteration and confirm the origins of syrups added to the honey.

“The large variation of honey composition makes it particularly difficult to authenticate. So having this consistent technique in the testing armoury could take the sting out of honey fraud.”

Sophie Dodd, who is completing her PhD on the topic of honey authentication at Cranfield University added, “It is vital to have samples of known origin and purity to validate the methods, so we want to extend our thanks to the Bee Farmers Association who we work closely with in our projects.”

The two methods developed can work together to increase chances of detecting exogenous sugar adulteration in honey.

Europeans and Native Americans valued their dogs as companion animals, using them for similar work and as symbols of identity. Consequently, the dogs reflected the tension between European and Indigenous cultures — the settlers described Indigenous dogs as mongrels to emphasize the perception that Indigenous people did not breed or own their dogs. Indigenous peoples identified European dogs as a direct threat to their existence and took measures to limit the use of European dogs.

“Previous studies had suggested that there were a lot of Indigenous dogs in the continental United States and that they were eradicated,” said Ariane Thomas, a recent PhD graduate of anthropology at the University of Iowa. “We wanted to understand what that entailed: when it happened, were they culled, was it the competition with European dogs, or was it disease?”

The researchers focused on the Jamestown colony in Virginia due to the number of canid remains available at the site and the evidence of Indigenous influence. They worked with Jamestown Rediscovery to identify and analyze 181 canid bones that represented at least 16 individual dogs. Of these, the team selected 22 remains that spanned multiple time points of the early settlement at Jamestown, between 1607 and 1619. They extracted the DNA at the ancient DNA lab in the Core Facilities of the Carl R. Woese Institute for Genomic Biology. The researchers then sequenced the data at the Roy J. Carver Biotechnology Center at Illinois to better understand the ancestry of these dogs.

“This project is a great example of the type of team science that we use at IGB, where people from diverse fields come together to answer questions through the use of complementary skill sets,” said Alida de Flamingh, a postdoctoral researcher in the Malhi (CIS/GSP/IGOH/GNDP) lab.

Based on body size estimates alone, the team discovered that most of the Jamestown dogs weighed between 22-39 lbs, comparable to modern-day beagles or schnauzers. Furthermore, many of the dog bones showed traces of human-inflicted damage, including burning and cut marks.

“The cut marks and other butchery marks we found on them show that some of these dogs were eaten. It implies that when the colonists came over, they didn’t have enough food and they had to rely on the Indigenous dogs in the area,” Thomas said.

Additionally, the DNA sequences demonstrated that at least six of the dogs showed evidence of Indigenous North American ancestry. “Our results show that there were Indigenous dogs in the area and they weren’t immediately eradicated when the Europeans arrived,” Thomas said.

Although the identification of dogs with Indigenous ancestry is not surprising, the results suggest that the colonists and Indigenous tribes may have traded dogs and likely had little concern with possible interbreeding. The researchers are interested in expanding to other sites and obtaining more high-quality DNA samples and reconstructions of dog body size to shed light on whether these dogs had full Indigenous ancestry or whether they were the product of mating with European dogs.

Traditionally, synthesizing solid-state materials has been a time-consuming and energy-intensive process, often accompanied by the production of harmful byproducts. But FWF enables gram-scale production of diverse compounds in seconds while reducing energy, water consumption and greenhouse gas emissions by more than 50%, setting a new standard for sustainable manufacturing.

The innovative research builds on Tour’s 2020 development of waste disposal and upcycling applications using flash Joule heating, a technique that passes a current through a moderately resistive material to quickly heat it to over 3,000 degrees Celsius (over 5,000 degrees Fahrenheit) and transform it into other substances.

“The key is that formerly we were flashing carbon and a few other compounds that could be conductive,” said Tour, the T.T. and W.F. Chao Professor of Chemistry and professor of materials science and nanoengineering. “Now we can flash synthesize the rest of the periodic table. It is a big advance.”

FWF’s success lies in its ability to overcome the conductivity limitations of conventional flash Joule heating methods. The team — including Ph.D. student Chi Hun “Will” Choi and corresponding author Yimo Han , assistant professor of chemistry, materials science and nanoengineering — incorporated an outer flash heating vessel filled with metallurgical coke and a semiclosed inner reactor containing the target reagents. FWF generates intense heat of about 2,000 degrees Celsius, which rapidly converts the reagents into high-quality materials through heat conduction.

This novel approach allows for the synthesis of more than 20 unique, phase-selective materials with high purity and consistency, according to the study. FWF’s versatility and scalability is ideal for the production of next-generation semiconductor materials such as molybdenum diselenide (MoSe2), tungsten diselenide and alpha phase indium selenide, which are notoriously difficult to synthesize using conventional techniques.

“Unlike traditional methods, FWF does not require the addition of conductive agents, reducing the formation of impurities and byproducts,” Choi said.

This advancement creates new opportunities in electronics, catalysis, energy and fundamental research. It also offers a sustainable solution for manufacturing a wide range of materials. Moreover, FWF has the potential to revolutionize industries such as aerospace, where materials like FWF-made MoSe2 demonstrate superior performance as solid-state lubricants.

“FWF represents a transformative shift in material synthesis,” Han said. “By providing a scalable and sustainable method for producing high-quality solid-state materials, it addresses barriers in manufacturing while paving the way for a cleaner and more efficient future.”