Computers Math

A new artificial intelligence-based score considers multiple factors to predict the prognosis of individual patients with COVID-19 seen at urgent care clinics or emergency departments. The tool, which was created by investigators at Massachusetts General Hospital (MGH), can be used to rapidly and automatically determine which patients are most likely to develop complications and need to be hospitalized.
The impetus for the study began early during the U.S. epidemic when Massachusetts experienced frequent urgent care visits and hospital admissions. While working as an infectious diseases physician and as part of the MGH Biothreats team, Gregory Robbins, MD, recognized the need for a more sophisticated method to identify outpatients at greatest risk for experiencing negative outcomes.
As described in The Journal of Infectious Diseases, a team of experts in neurology, infectious disease, critical care, radiology, pathology, emergency medicine and machine learning designed the COVID-19 Acuity Score (CoVA) based on input from information on 9,381 adult outpatients seen in MGH’s respiratory illness clinics and emergency department between March 7 and May 2, 2020. “The large sample size helped ensure that the machine learning model was able to learn which of the many different pieces of data available allow reliable predictions about the course of COVID-19 infection,” said M. Brandon Westover, MD, PhD, an investigator in the Department of Neurology and director of Data Science at the MGH McCance Center for Brain Health. Westover is one of three co-senior authors of the study, along with Robbins and Shibani Mukerji, MD, PhD, associate director of MGH’s Neuro-Infectious Diseases Unit.
CoVA was then tested in another 2,205 patients seen between May 3 and May 14. “Testing the model prospectively helped us to verify that the CoVA score actually works when it sees ‘new’ patients, in the real world,” said first author Haoqi Sun, PhD, an investigator in the Department of Neurology and a research faculty member in the MGH Clinical Data Animation Center (CDAC). In this prospective validation group, 26.1 percent, 6.3 percent and 0.5 percent of patients experienced hospitalization, critical illness or death, respectively, within seven days. CoVA demonstrated excellent performance in predicting which patients would fall into these categories.
Among 30 predictors — which included demographics like age and gender, COVID-19 testing status, vital signs, medical history and chest X-ray results (when available) — the top five were age, diastolic blood pressure, blood oxygen saturation, COVID-19 testing status and respiratory rate.
“While several other groups have developed risk scores for complications of COVID-19, ours is unique in being based on such a large patient sample, being prospectively validated, and in being specifically designed for use in the outpatient setting, rather than for patients who are already hospitalized,” Mukerji said. “CoVA is designed so that automated scoring could be incorporated into electronic medical record systems. We hope that it will be useful in case of future COVID-19 surges, when rapid clinical assessments may be critical.”

Story Source:
Materials provided by Massachusetts General Hospital. Note: Content may be edited for style and length. More

With a training technique commonly used to teach dogs to sit and stay, Johns Hopkins University computer scientists showed a robot how to teach itself several new tricks, including stacking blocks. With the method, the robot, named Spot, was able to learn in days what typically takes a month.
By using positive reinforcement, an approach familiar to anyone who’s used treats to change a dog’s behavior, the team dramatically improved the robot’s skills and did it quickly enough to make training robots for real-world work a more feasible enterprise. The findings are newly published in a paper called, “Good Robot!”
“The question here was how do we get the robot to learn a skill?” said lead author Andrew Hundt, a PhD student working in Johns Hopkins’ Computational Interaction and Robotics Laboratory. “I’ve had dogs so I know rewards work and that was the inspiration for how I designed the learning algorithm.”
Unlike humans and animals that are born with highly intuitive brains, computers are blank slates and must learn everything from scratch. But true learning is often accomplished with trial and error, and roboticists are still figuring out how robots can learn efficiently from their mistakes.
The team accomplished that here by devising a reward system that works for a robot the way treats work for a dog. Where a dog might get a cookie for a job well done, the robot earned numeric points.
Hundt recalled how he once taught his terrier mix puppy named Leah the command “leave it,” so she could ignore squirrels on walks. He used two types of treats, ordinary trainer treats and something even better, like cheese. When Leah was excited and sniffing around the treats, she got nothing. But when she calmed down and looked away, she got the good stuff. “That’s when I gave her the cheese and said, ‘Leave it! Good Leah!'”
Similarly, to stack blocks, Spot the robot needed to learn how to focus on constructive actions. As the robot explored the blocks, it quickly learned that correct behaviors for stacking earned high points, but incorrect ones earned nothing. Reach out but don’t grasp a block? No points. Knock over a stack? Definitely no points. Spot earned the most by placing the last block on top of a four-block stack.

advertisement

The training tactic not only worked, it took just days to teach the robot what used to take weeks. The team was able to reduce the practice time by first training a simulated robot, which is a lot like a video game, then running tests with Spot.
“The robot wants the higher score,” Hundt said. “It quickly learns the right behavior to get the best reward. In fact, it used to take a month of practice for the robot to achieve 100% accuracy. We were able to do it in two days.”
Positive reinforcement not only worked to help the robot teach itself to stack blocks, with the point system the robot just as quickly learned several other tasks — even how to play a simulated navigation game. The ability to learn from mistakes in all types of situations is critical for designing a robot that could adapt to new environments.
“At the start the robot has no idea what it’s doing but it will get better and better with each practice. It never gives up and keeps trying to stack and is able to finish the task 100% of the time,” Hundt said.
The team imagines these findings could help train household robots to do laundry and wash dishes — tasks that could be popular on the open market and help seniors live independently. It could also help design improved self-driving cars.
“Our goal is to eventually develop robots that can do complex tasks in the real world — like product assembly, caring for the elderly and surgery,” Hager said. “We don’t currently know how to program tasks like that — the world is too complex. But work like this shows us that there is promise to the idea that robots can learn how to accomplish such real-world tasks in a safe and efficient way.”

Story Source:
Materials provided by Johns Hopkins University. Note: Content may be edited for style and length. More

Engineers at Rice University and Texas A&M University have found a 2D material that could make computers faster and more energy-efficient.
Their material is a derivative of perovskite — a crystal with a distinctive structure — that has the surprising ability to enable the valleytronics phenomenon touted as a possible platform for information processing and storage.
The lab of materials scientist Jun Lou of Rice’s Brown School of Engineering synthesized a layered compound of cesium, bismuth and iodine that is adept at storing the valley states of electrons, but only in the structure’s odd layers.
These bits can be set with polarized light, and the even layers appear to protect the odd ones from the kind of field interference that bedevils other perovskites, according to the researchers.
Best of all, the material appears to be scalable.
“This is not a new material, but we figured out a way to make it without solution processing or exfoliating it from bulk,” Lou said. “What’s novel is that we can produce it (via chemical vapor deposition) in a few layers, and all the way down to a monolayer. That enabled us to probe its nonlinear optical properties.”
The discovery is detailed in Advanced Materials.

advertisement

Valleytronics are a cousin to spintronics, in which memory bits are defined by an electron’s quantum spin state. In valleytronics, electrons have degrees of freedom in the multiple momentum states — or valleys — they occupy. These states can be read as bits.
“In a transistor, if you put an electron there, it represents a state, and if you take it out, that represents another state,” said co-principal investigator Hanyu Zhu of Rice. “In valleytronics, the electrons are always present, and are in either of two different quantum wavefunctions with opposite momenta. These two wavefunctions interact with different light polarization, so the momentum state can be resolved optically.”
A close look at the inorganic, lead-free material through an electron microscope showed molecules in the odd layer are asymmetric. “That lack of symmetry is missing in the even layers — that’s how we differentiate between them — and it gives rise to the properties we see,” Lou said. “That’s just the nature of this crystal structure.”
The lab tested the material with up to 11 layers and found a lack of transparency doesn’t seem to affect how well light triggered a response. “Even a thicker material behaves like it’s still a single layer,” Lou said. “That’s quite important.”
“Thicker 2D transition metal dichalcogenides lose unique properties like valleytronics,” he said. “All the behaviors are gone. That’s not the case for this material.”
Lou said calculations by co-principal investigator Xiaofeng Qian of Texas A&M University provided the necessary theoretical evidence.

advertisement

“The valley polarization observed in both thin and thick layers is largely due to the weak interlayer electronic coupling, a unique feature of this perovskite derivative compared to other 2D materials when stacked together,” Qian said. “It also leads to persistent nonlinear optical responses in thicker samples.”
The material also seems less susceptible to environmental degradation, a common problem for hybrid perovskites developed for solar energy. “This material won’t give you very high conversion efficiency, but think of it like an all-around athlete in the Olympic Games,” said lead author and Rice postdoctoral fellow Jia Liang. “It may not be the best in each category, but if you consider its different aspects together, it will stand out,” he said.
The researchers suggested the already strong light-matter interaction they observed could be enhanced by further engineering the material’s band gap.
“I think it’s a breakthrough for using this type of material in information processing,” Lou said. “We’re really hoping this is the starting point.” More

A University of Central Florida researcher is working to make portable devices and electric vehicles stay charged longer by extending the life of the rechargeable lithium-ion batteries powering them.
Assistant Professor Yang Yang is doing this by making the batteries more efficient, with some of his latest work focusing on keeping an internal metal structure, the anode, from falling apart over time by applying a thin, film-like coating of copper and tin. The new technique is detailed in a recent study in the journal Advanced Materials.
An anode generates electrons that travel to a similar structure, the cathode, inside the battery, thus creating a current and power.
“Our work has shown that the rate of degradation of the anode can be reduced by more than 1,000 percent by using a copper-tin film compared to a tin film that is often used,” said Yang, who is with UCF’s NanoScience Technology Center.
Yang is an expert in battery improvement including making them safer and able to withstand extreme temperatures.
The technique is unique because of its use of the copper-tin alloy and is an important improvement in stabilizing rechargeable battery performance, Yang says.
It is also scalable for use in the smallest smartphone battery to larger batteries that power electric cars and trucks.
“We are motivated by our most recent research progress in alloyed materials for various applications,” he says. “Each alloy is unique in composition, structure and property.”

Story Source:
Materials provided by University of Central Florida. Original written by Robert Wells. Note: Content may be edited for style and length. More

Maths — it’s the subject some kids love to hate, yet despite its lack of popularity, mathematics is critical for a STEM-capable workforce and vital for Australia’s current and future productivity.
In a new study by the University of South Australia in collaboration with the Australian Council for Educational Research, researchers have been exploring the impact of anxiety on learning maths, finding that boosting student confidence, is pivotal to greater engagement with the subject.
Maths anxiety, or ‘mathemaphobia’, is the sense of fear, worry and nervousness that students may experience when participating in mathematical tasks.
In Australia a quarter to a third of Australian secondary students report feeling tense, nervous or helpless when doing maths, and it’s this reaction that’s influencing their decisions to study maths.
Lead researcher, Dr Florence Gabriel says maths anxiety is one of the biggest barriers to students choosing to study it, especially at senior school and tertiary levels.
“Many of us would have felt some sort of maths anxiety in the past — a sense of panic or worry, feelings of failure, or even a faster heart rate — all of which are associated with stress,” Dr Gabriel says.

advertisement

“Maths anxiety is essentially an emotional reaction, but it’s just like stress in other situations.
“When students experience maths anxiety, they’ll tend to hurry through maths questions, lose focus, or simply give up when it all seems too hard. Not surprisingly, these reactions compound and lead to poor maths achievement — and later a reluctance to engage with the subject at all.
“To break this cycle, our research shows that we need to build and grow student confidence in maths, especially before starting a new maths concept.
“This draws on the notion of self-regulated learning ¬- where students have the ability to understand, track and control their own learning.
“By drawing a student’s attention to instances where they’ve previously overcome a difficult maths challenge, or to a significant maths success, we’re essentially building their confidence and belief in their own abilities, and it’s this that will start to counteract negative emotions.”
The study assessed the responses of 4295 Australian 15-year-old students that participated in the 2012 cycle of the OECD’s Program for International Student Assessment (PISA).

advertisement

It focussed on the psychological factors of maths learning: motivation (the belief that maths is important and useful for their future); maths self-concept (the belief in their ability to do maths); maths anxiety (self-feelings when doing maths); perseverance (their willingness to continue to work on difficult problems); maths self-efficacy (their self-belief that they can successfully solve maths problems); and maths literacy (the ability to apply maths to the real world).
“Importantly, our research shows the domino effect that these variables have on one another,” Dr Gabriel says.
“Through structural equation modelling, our data shows that low motivation and self-concept will lead to maths anxiety, which in turn affects perseverance, self-efficacy and, ultimately, maths achievement.
“By developing a student’s ability to reflect on past successes — before maths anxiety sets in — we can break through some of the negative and emotional beliefs about maths and, hopefully, pave the way for students to accept and engage with maths in the future.” More

Kids with wildly popular YouTube channels are frequently promoting unhealthy food and drinks in their videos, warn researchers at NYU School of Global Public Health and NYU Grossman School of Medicine in a new study published in the journal Pediatrics.
Food and beverage companies spend $1.8 billion dollars a year marketing their products to young people. Although television advertising is a major source of food marketing, companies have dramatically increased online advertising in response to consumers’ growing social media use.
“Kids already see several thousand food commercials on television every year, and adding these YouTube videos on top of it may make it even more difficult for parents and children to maintain a healthy diet,” said Marie Bragg, assistant professor of public health nutrition at NYU School of Global Public Health and assistant professor in the Department of Population Health at NYU Langone. “We need a digital media environment that supports healthy eating instead of discouraging it.”
YouTube is the second most visited website in the world and is a popular destination for kids seeking entertainment. More than 80 percent of parents with a child younger than 12 years old allow their child to watch YouTube, and 35 percent of parents report that their kid watches YouTube regularly.
“The allure of YouTube may be especially strong in 2020 as many parents are working remotely and have to juggle the challenging task of having young kids at home because of COVID-19,” said Bragg, the study’s senior author.
When finding videos for young children to watch, millions of parents turn to videos of “kid influencers,” or children whose parents film them doing activities such as science experiments, playing with toys, or celebrating their birthdays. The growing popularity of these YouTube videos have caught the attention of companies, who advertise or sponsor posts to promote their products before or during videos. In fact, the highest-paid YouTube influencer of the past two years was an 8-year-old who earned $26 million last year.

advertisement

“Parents may not realize that kid influencers are often paid by food companies to promote unhealthy food and beverages in their videos. Our study is the first to quantify the extent to which junk food product placements appear in YouTube videos from kid influencers,” said Bragg.
Bragg and her colleagues identified the five most popular kid influencers on YouTube of 2019 — whose ages ranged from 3 to 14 years old — and analyzed their most-watched videos. Focusing on a sample of 418 YouTube videos, they recorded whether food or drinks were shown in the videos, what items and brands were shown, and assessed their nutritional quality.
The researchers found that nearly half of the most-popular videos from kid influencers (42.8 percent) promoted food and drinks. More than 90 percent of the products shown were unhealthy branded food, drinks, or fast food toys, with fast food as the most frequently featured junk food, followed by candy and soda. Only a few videos featured unhealthy unbranded items like hot dogs (4 percent), healthy unbranded items like fruit (3 percent), and healthy branded items like yogurt brands (2 percent).
The videos featuring junk food product placements were viewed more than 1 billion times — a staggering level of exposure for food and beverage companies.
“It was concerning to see that kid influencers are promoting a high volume of junk food in their YouTube videos, and that those videos are generating enormous amounts of screen time for these unhealthy products,” said Bragg.
While the researchers do not know which food and drink product placements were paid endorsements, they find these videos problematic for public health because they enable food companies to directly — but subtly — promote unhealthy foods to young children and their parents.
“It’s a perfect storm for encouraging poor nutrition — research shows that people trust influencers because they appear to be ‘everyday people,’ and when you see these kid influencers eating certain foods, it doesn’t necessarily look like advertising. But it is advertising, and numerous studies have shown that children who see food ads consume more calories than children who see non-food ads, which is why the National Academy of Medicine and World Health Organization identify food marketing as a major driver of childhood obesity,” said Bragg.
The researchers encourage federal and state regulators to strengthen and enforce regulations of junk food advertising by kid influencers.
“We hope that the results of this study encourage the Federal Trade Commission and state attorneys general to focus on this issue and identify strategies to protect children and public health,” said study co-author Jennifer Pomeranz, assistant professor of public health policy and management at NYU School of Global Public Health. More

Extreme events occur in many observable contexts. Nature is a prolific source: rogue water waves surging high above the swell, monsoon rains, wildfire, etc. From climate science to optics, physicists have classified the characteristics of extreme events, extending the notion to their respective domains of expertise. For instance, extreme events can take place in telecommunication data streams. In fiber-optic communications where a vast number of spatio-temporal fluctuations can occur in transoceanic systems, a sudden surge is an extreme event that must be suppressed, as it can potentially alter components associated with the physical layer or disrupt the transmission of private messages.
Recently, extreme events have been observed in quantum cascade lasers, as reported by researchers from Télécom Paris (France) in collaboration with UC Los Angeles (USA) and TU Darmstad (Germany). The giant pulses that characterize these extreme events can contribute the sudden, sharp bursts necessary for communication in neuromorphic systems inspired by the brain’s powerful computational abilities. Based on a quantum cascade laser (QCL) emitting mid-infrared light, the researchers developed a basic optical neuron system operating 10,000× faster than biological neurons. Their report is published in Advanced Photonics.
Giant pulses, fine tuning
Olivier Spitz, Télécom Paris research fellow and first author on the paper, notes that the giant pulses in QCLs can be triggered successfully by adding a “pulse-up excitation,” a short-time small-amplitude increase of bias current. Senior author Frédéric Grillot, Professor at Télécom Paris and the University of New Mexico, explains that this triggering ability is of paramount importance for applications such as optical neuron-like systems, which require optical bursts to be triggered in response to a perturbation.
The team’s optical neuron system demonstrates behaviors like those observed in biological neurons, such as thresholding, phasic spiking, and tonic spiking. Fine tuning of modulation and frequency allows control of time intervals between spikes. Grillot explains, “The neuromorphic system requires a strong, super-threshold stimulus for the system to fire a spiking response, whereas phasic and tonic spiking correspond to single or continuous spike firing following the arrival of a stimulus.” To replicate the various biological neuronal responses, interruption of regular successions of bursts corresponding to neuronal activity is also required.
Quantum cascade laser
Grillot notes that the findings reported by his team demonstrate the increasingly superior potential of quantum cascade lasers compared to standard diode lasers or VCSELs, for which more complex techniques are currently required to achieve neuromorphic properties.
Experimentally demonstrated for the first time in 1994, quantum cascade lasers were originally developed for use under cryogenic temperatures. Their development has advanced rapidly, allowing use at warmer temperatures, up to room temperature. Due to the large number of wavelengths they can achieve (from 3 to 300 microns), QCLs contribute to many industrial applications such as spectroscopy, optical countermeasures, and free-space communications.
According to Grillot, the physics involved in QCLs is totally different than that in diode lasers. “The advantage of quantum cascade lasers over diode lasers comes from the sub-picosecond electronic transitions among the conduction-band states (subbands) and a carrier lifetime much shorter than the photon lifetime,” says Grillot. He remarks that QCLs exhibit completely different light emission behaviors under optical feedback, including but not limited to giant pulse occurrences, laser responses to modulation, and frequency comb dynamics.

Story Source:
Materials provided by SPIE–International Society for Optics and Photonics. Original written by Renae Keep. Note: Content may be edited for style and length. More

By expanding on existing designs for electrodes of ultra-thin solar panels, Stanford researchers and collaborators in Korea have developed a new architecture for OLED — organic light-emitting diode — displays that could enable televisions, smartphones and virtual or augmented reality devices with resolutions of up to 10,000 pixels per inch (PPI). (For comparison, the resolutions of new smartphones are around 400 to 500 PPI.)
Such high-pixel-density displays will be able to provide stunning images with true-to-life detail — something that will be even more important for headset displays designed to sit just centimeters from our faces.
The advance is based on research by Stanford University materials scientist Mark Brongersma in collaboration with the Samsung Advanced Institute of Technology (SAIT). Brongersma was initially put on this research path because he wanted to create an ultra-thin solar panel design.
“We’ve taken advantage of the fact that, on the nanoscale, light can flow around objects like water,” said Brongersma, who is a professor of materials science and engineering and senior author of the Oct. 22 Science paper detailing this research. “The field of nanoscale photonics keeps bringing new surprises and now we’re starting to impact real technologies. Our designs worked really well for solar cells and now we have a chance to impact next generation displays.”
In addition to having a record-setting pixel density the new “metaphotonic” OLED displays would also be brighter and have better color accuracy than existing versions, and they’d be much easier and cost-effective to produce as well.
Hidden gems
At the heart of an OLED are organic, light-emitting materials. These are sandwiched between highly-reflective and semi-transparent electrodes that enable current injection into the device. When electricity flows through an OLED, the emitters give off red, green or blue light. Each pixel in an OLED display is composed of smaller sub-pixels that produce these primary colors. When the resolution is sufficiently high, the pixels are perceived as one color by the human eye. OLEDs are an attractive technology because they are thin, light and flexible and produce brighter and more colorful images than other kinds of displays.

advertisement

This research aims to offer an alternative to the two types of OLED displays that are currently commercially available. One type — called a red-green-blue OLED — has individual sub-pixels that each contain only one color of emitter. These OLEDs are fabricated by spraying each layer of materials through a fine metal mesh to control the composition of each pixel. They can only be produced on a small scale, however, like what would be used for a smartphone.
Larger devices like TVs employ white OLED displays. Each of these sub-pixels contains a stack of all three emitters and then relies on filters to determine the final sub-pixel color, which is simpler to fabricate. Since the filters reduce the overall output of light, white OLED displays are more power-hungry and prone to having images burn into the screen.
OLED displays were on the mind of Won-Jae Joo, a SAIT scientist, when he visited Stanford from 2016 to 2018. During that time, Joo listened to a presentation by Stanford graduate student Majid Esfandyarpour about an ultrathin solar cell technology he was developing in Brongersma’s lab and realized it had applications beyond renewable energy.
“Professor Brongersma’s research themes were all very academically profound and were like hidden gems for me as an engineer and researcher at Samsung Electronics,” said Joo, who is lead author of the Science paper.
Joo approached Esfandyarpour after the presentation with his idea, which led to a collaboration between researchers at Stanford, SAI and Hanyang University in Korea.

advertisement

“It was quite exciting to see that a problem that we have already thought about in a different context can have such an important impact on OLED displays,” said Esfandyarpour.
A fundamental foundation
The crucial innovation behind both the solar panel and the new OLED is a base layer of reflective metal with nanoscale (smaller than microscopic) corrugations, called an optical metasurface. The metasurface can manipulate the reflective properties of light and thereby allow the different colors to resonate in the pixels. These resonances are key to facilitating effective light extraction from the OLEDs.
“This is akin to the way musical instruments use acoustic resonances to produce beautiful and easily audible tones,” said Brongersma, who conducted this research as part of the Geballe Laboratory for Advanced Materials at Stanford.
For example, red emitters have a longer wavelength of light than blue emitters, which, in conventional RGB-OLEDs, translates to sub-pixels of different heights. In order to create a flat screen overall, the materials deposited above the emitters have to be laid down in unequal thicknesses. By contrast, in the proposed OLEDs, the base layer corrugations allow each pixel to be the same height and this facilitates a simpler process for large-scale as well as micro-scale fabrication.
In lab tests, the researchers successfully produced miniature proof-of-concept pixels. Compared with color-filtered white-OLEDs (which are used in OLED televisions) these pixels had a higher color purity and a twofold increase in luminescence efficiency — a measure of how bright the screen is compared to how much energy it uses. They also allow for an ultrahigh pixel density of 10,000 pixels-per-inch.
The next steps for integrating this work into a full-size display is being pursued by Samsung, and Brongersma eagerly awaits the results, hoping to be among the first people to see the meta-OLED display in action. More