Computers Math

From swallowing pills to injecting insulin, patients frequently administer their own medication. But they don’t always get it right. Improper adherence to doctors’ orders is commonplace, accounting for thousands of deaths and billions of dollars in medical costs annually. MIT researchers have developed a system to reduce those numbers for some types of medications.
The new technology pairs wireless sensing with artificial intelligence to determine when a patient is using an insulin pen or inhaler, and flags potential errors in the patient’s administration method. “Some past work reports that up to 70% of patients do not take their insulin as prescribed, and many patients do not use inhalers properly,” says Dina Katabi, the Andrew and Erna Viteri Professor at MIT, whose research group has developed the new solution. The researchers say the system, which can be installed in a home, could alert patients and caregivers to medication errors and potentially reduce unnecessary hospital visits.
The research appears in the journal Nature Medicine. The study’s lead authors are Mingmin Zhao, a PhD student in MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL), and Kreshnik Hoti, a former visiting scientist at MIT and current faculty member at the University of Prishtina in Kosovo. Other co-authors include Hao Wang, a former CSAIL postdoc and current faculty member at Rutgers University, Aniruddh Raghu, a CSAIL PhD student.
Some common drugs entail intricate delivery mechanisms. “For example, insulin pens require priming to make sure there are no air bubbles inside. And after injection, you have to hold for 10 seconds,” says Zhao. “All those little steps are necessary to properly deliver the drug to its active site.” Each step also presents opportunity for errors, especially when there’s no pharmacist present to offer corrective tips. Patients might not even realize when they make a mistake — so Zhao’s team designed an automated system that can.
Their system can be broken down into three broad steps. First, a sensor tracks a patient’s movements within a 10-meter radius, using radio waves that reflect off their body. Next, artificial intelligence scours the reflected signals for signs of a patient self-administering an inhaler or insulin pen. Finally, the system alerts the patient or their health care provider when it detects an error in the patient’s self-administration.
The researchers adapted their sensing method from a wireless technology they’d previously used to monitor people’s sleeping positions. It starts with a wall-mounted device that emits very low-power radio waves. When someone moves, they modulate the signal and reflect it back to the device’s sensor. Each unique movement yields a corresponding pattern of modulated radio waves that the device can decode. “One nice thing about this system is that it doesn’t require the patient to wear any sensors,” says Zhao. “It can even work through occlusions, similar to how you can access your Wi-Fi when you’re in a different room from your router.”
The new sensor sits in the background at home, like a Wi-Fi router, and uses artificial intelligence to interpret the modulated radio waves. The team developed a neural network to key in on patterns indicating the use of an inhaler or insulin pen. They trained the network to learn those patterns by performing example movements, some relevant (e.g. using an inhaler) and some not (e.g. eating). Through repetition and reinforcement, the network successfully detected 96 percent of insulin pen administrations and 99 percent of inhaler uses.
Once it mastered the art of detection, the network also proved useful for correction. Every proper medicine administration follows a similar sequence — picking up the insulin pen, priming it, injecting, etc. So, the system can flag anomalies in any particular step. For example, the network can recognize if a patient holds down their insulin pen for five seconds instead of the prescribed 10 seconds. The system can then relay that information to the patient or directly to their doctor, so they can fix their technique.
“By breaking it down into these steps, we can not only see how frequently the patient is using their device, but also assess their administration technique to see how well they’re doing,” says Zhao.
The researchers say a key feature of their radio wave-based system is its noninvasiveness. “An alternative way to solve this problem is by installing cameras,” says Zhao. “But using a wireless signal is much less intrusive. It doesn’t show peoples’ appearance.”
He adds that their framework could be adapted to medications beyond inhalers and insulin pens — all it would take is retraining the neural network to recognize the appropriate sequence of movements. Zhao says that “with this type of sensing technology at home, we could detect issues early on, so the person can see a doctor before the problem is exacerbated.” More

Training neural networks to perform tasks, such as recognizing images or navigating self-driving cars, could one day require less computing power and hardware thanks to a new artificial neuron device developed by researchers at the University of California San Diego. The device can run neural network computations using 100 to 1000 times less energy and area than existing CMOS-based hardware.
Researchers report their work in a paper published Mar. 18 in Nature Nanotechnology.
Neural networks are a series of connected layers of artificial neurons, where the output of one layer provides the input to the next. Generating that input is done by applying a mathematical calculation called a non-linear activation function. This is a critical part of running a neural network. But applying this function requires a lot of computing power and circuitry because it involves transferring data back and forth between two separate units — the memory and an external processor.
Now, UC San Diego researchers have developed a nanometer-sized device that can efficiently carry out the activation function.
“Neural network computations in hardware get increasingly inefficient as the neural network models get larger and more complex,” said Duygu Kuzum, a professor of electrical and computer engineering at the UC San Diego Jacobs School of Engineering. “We developed a single nanoscale artificial neuron device that implements these computations in hardware in a very area- and energy-efficient way.”
The new study, led by Kuzum and her Ph.D. student Sangheon Oh, was performed in collaboration with a DOE Energy Frontier Research Center led by UC San Diego physics professor Ivan Schuller, which focuses on developing hardware implementations of energy-efficient artificial neural networks. More

If you’re looking for one technique to maximize photon output from plasmons, stop. It takes two to wrangle.
Rice University physicists came across a phenomenon that boosts the light from a nanoscale device more than 1,000 times greater than they anticipated.
When looking at light coming from a plasmonic junction, a microscopic gap between two gold nanowires, there are conditions in which applying optical or electrical energy individually prompted only a modest amount of light emission. Applying both together, however, caused a burst of light that far exceeded the output under either individual stimulus.
The researchers led by Rice physicist Douglas Natelson and lead authors Longji Cui and Yunxuan Zhu found the effect while following up experiments that discovered driving current through the gap increased the number of light-emitting “hot carrier” electrons in the electrodes.
Now they know that adding energy from a laser to the same junction makes it even brighter. The effect could be employed to make nanophotonic switches for computer chips and for advanced photocatalysts.
The details appear in the American Chemical Society journal Nano Letters.
“It’s been known for a long time that it’s possible to get a light emission from these tiny structures,” Natelson said. “In our previous work, we showed that plasmons play an important role in generating very hot charge carriers, equivalent to a couple of thousand degrees.” More

In their pursuit of maximum reward, people suffering from gambling disorder rely less on exploring new but potentially better strategies, and more on proven courses of action that have already led to success in the past. The neurotransmitter dopamine in the brain may play an important role in this, a study in biological psychology conducted at the University of Cologne’s Faculty of Human Sciences by Professor Dr Jan Peters und Dr Antonius Wiehler suspects. The article ‘Attenuated directed exploration during reinforcement learning in gambling disorder’ has appeared in the latest edition of the Journal of Neuroscience, published by the Society for Neuroscience.
Gambling disorder affects slightly less than one percent of the population — often men — and is in some ways similar to substance abuse disorders. Scientists suspect that this disorder, like other addiction disorders, is associated with changes in the dopamine system. The brain’s reward system releases the neurotransmitter dopamine during gambling. Since dopamine is important for the planning and control of actions, among other things, it could also affect strategic learning processes.
‘Gambling disorder is of scientific interest among other things because it is an addiction disorder that is not tied to a specific substance’, Professor Dr Jan Peters, one of the authors, remarked. The psychologists examined how gamblers plan their actions to maximize rewards — how their so called reinforcement learning works. In the study, participants had to decide between already proven options or new ones in order to win as much as possible. At the same time, the scientists used functional magnetic resonance imaging to measure activity in regions of the brain that are important for processing reward stimuli and planning actions.
Twenty-three habitual gamblers and twenty-three control subjects (all male) performed what is known as a ‘four-armed bandit task’. The name of this type of decision-making task refers to slot machines, known colloquially as ‘one-armed bandits’. In each run, the participants had to choose between four options (‘four-armed bandit’, in this case four coloured squares), whose winnings slowly changed. Different strategies can be employed here. For example, one can choose the option that yielded the highest profit last time. However, it is also possible to choose the option where the chance of winning is most uncertain — the option promising maximum information gain. The latter is also called directed (or uncertainty-based) exploration.
Both groups won about the same amount of money and exhibited directed exploration. However, this was significantly less pronounced in the group of gamblers than in the control group. These results indicate that gamblers are less adaptive to changing environments during reinforcement learning. At the neural level, gamblers showed changes in a network of brain regions that has been associated with directed exploration in previous studies. In one previous study by the two biological psychologists, pharmacologically raising the dopamine level in healthy participants had shown a very similar effect on behaviour. ‘Although this indicates that dopamine might also play an important role in the reduction of directed exploration in gamblers, more research would have to be conducted to prove such a correlation,’ said Dr Antonius Wiehler.
Further research also needs to clarify whether the observed changes in decision-making behaviour in gamblers are a risk factor for, or a consequence of, regular gambling.
Story Source:
Materials provided by University of Cologne. Note: Content may be edited for style and length. More

Using your face to unlock your phone is a pretty genius security protocol. But like any advanced technology, hackers and thieves are always up to the challenge, whether that’s unlocking your phone with your face while you sleep or using a photo from social media to do the same.
Like every other human biometric identification system before it (fingerprints, retina scans) there are still significant security flaws in some of the most advanced identity verification technology. Brigham Young University electrical and computer engineering professor D.J. Lee has decided there is a better and more secure way to use your face for restricted access.
It’s called Concurrent Two-Factor Identity Verification (C2FIV) and it requires both one’s facial identity and a specific facial motion to gain access. To set it up, a user faces a camera and records a short 1-2 second video of either a unique facial motion or a lip movement from reading a secret phrase. The video is then input into the device, which extracts facial features and the features of the facial motion, storing them for later ID verification.
“The biggest problem we are trying to solve is to make sure the identity verification process is intentional,” said Lee, a professor of electrical and computer engineering at BYU. “If someone is unconscious, you can still use their finger to unlock a phone and get access to their device or you can scan their retina. You see this a lot in the movies — think of Ethan Hunt in Mission Impossible even using masks to replicate someone else’s face.”
To get technical, C2FIV relies on an integrated neural network framework to learn facial features and actions concurrently. This framework models dynamic, sequential data like facial motions, where all the frames in a recording have to be considered (unlike a static photo with a figure that can be outlined).
Using this integrated neural network framework, the user’s facial features and movements are embedded and stored on a server or in an embedded device and when they later attempt to gain access, the computer compares the newly generated embedding to the stored one. That user’s ID is verified if the new and stored embeddings match at a certain threshold.
“We’re pretty excited with the technology because it’s pretty unique to add another level of protection that doesn’t cause more trouble for the user,” Lee said.
In their preliminary study, Lee and his Ph.D. student Zheng Sun recorded 8,000 video clips from 50 subjects making facial movements such as blinking, dropping their jaw, smiling or raising their eyebrows as well as many random facial motions to train the neural network. They then created a dataset of positive and negative pairs of facial motions and inputted higher scores for the positive pairs (those that matched). Currently, with the small dataset, the trained neural network verifies identities with over 90% accuracy. They are confident the accuracy can be much higher with a larger dataset and improvements on the network.
Lee, who has filed a patent on the tech already, said the idea is not to compete with Apple or have the application be all about smartphone access. In his opinion, C2FIV has broader application, including accessing restricted areas at a workplace, online banking, ATM use, safe deposit box access or even hotel room entry or keyless entry/access to your vehicle.
“We could build this very tiny device with a camera on it and this device could be deployed easily at so many different locations,” Lee said. “How great would it be to know that even if you lost your car key, no one can steal your vehicle because they don’t know your secret facial action?”
Story Source:
Materials provided by Brigham Young University. Original written by Todd Hollingshead. Note: Content may be edited for style and length. More

If you want to build a fully functional nanosized robot, you need to incorporate a host of capabilities, from complicated electronic circuits and photovoltaics to sensors and antennas.
But just as importantly, if you want your robot to move, you need it to be able to bend.
Cornell researchers have created micron-sized shape memory actuators that enable atomically thin two-dimensional materials to fold themselves into 3D configurations. All they require is a quick jolt of voltage. And once the material is bent, it holds its shape — even after the voltage is removed.
As a demonstration, the team created what is potentially the world’s smallest self-folding origami bird. And it’s not a lark.
The group’s paper, “Micrometer-sized electrically programmable shape memory actuators for low-power microrobotics,” published in Science Robotics and was featured on the cover. The paper’s lead author is postdoctoral researcher Qingkun Liu.
The project is led by Itai Cohen, professor of physics, and Paul McEuen, the John A. Newman Professor of Physical Science. More

In researching the causes and potential treatments for degenerative conditions such as Alzheimer’s or Parkinson’s disease, neuroscientists frequently struggle to accurately identify cells needed to understand brain activity that gives rise to behavior changes such as declining memory or impaired balance and tremors.
A multidisciplinary team of Georgia Institute of Technology neuroscience researchers, borrowing from existing tools such as graphical models, have uncovered a better way to identify cells and understand the mechanisms of the diseases, potentially leading to better understanding, diagnosis, and treatment.
Their research findings were reported Feb. 24 in the journal eLife. The research was supported by the National Institutes of Health and the National Science Foundation.
The field of neuroscience studies how the nervous system functions, and how genes and environment influence behavior. By using new technologies to understand natural and dysfunctional states of biological systems, neuroscientists hope to ultimately bring cures to diseases. Before that can happen, neuroscientists first must understand which cells in the brain are driving behavior but mapping the brain activity cell by cell isn’t as simple as it appears.
No Two Brain Cells Are Alike
Traditionally, scientists established a coordinate system to map each cell location by comparing images to an atlas, but the notion in literature that “all brains look the same is absolutely not true,” said Hang Lu, the Love Family Professor of Chemical and Biomolecular Engineering in Georgia Tech’s School of Chemical and Biomolecular Engineering. More

A sustainable, powerful micro-supercapacitor may be on the horizon, thanks to an international collaboration of researchers from Penn State and the University of Electronic Science and Technology of China. Until now, the high-capacity, fast-charging energy storage devices have been limited by the composition of their electrodes — the connections responsible for managing the flow of electrons during charging and dispensing energy. Now, researchers have developed a better material to improve connectivity while maintaining recyclability and low cost.
They published their results on Feb. 8 in the Journal of Materials Chemistry A.
“The supercapacitor is a very powerful, energy-dense device with a fast-charging rate, in contrast to the typical battery — but can we make it more powerful, faster and with a really high retention cycle?” asked Jia Zhu, corresponding author and doctoral student conducting research in the laboratory of Huanyu “Larry” Cheng, Dorothy Quiggle Career Development Professor in Penn State’s Department of Engineering Science and Mechanics.
Zhu worked under Cheng’s mentorship to explore the connections in a micro-supercapacitor, which they use in their research on small, wearable sensors to monitor vital signs and more. Cobalt oxide, an abundant, inexpensive material that has a theoretically high capacity to quickly transfer energy charges, typically makes up the electrodes. However, the materials that mix with cobalt oxide to make an electrode can react poorly, resulting in a much lower energy capacity than theoretically possible.
The researchers ran simulations of materials from an atomic library to see if adding another material — also called doping — could amplify the desired characteristics of cobalt oxide as an electrode by providing extra electrons while minimizing, or entirely removing, the negative effects. They modeled various material species and levels to see how they would interact with cobalt oxide.
“We screened possible materials but found many that might work were too expensive or toxic, so we selected tin,” Zhu said. “Tin is widely available at a low cost, and it’s not harmful to the environment.”
In the simulations, the researchers found that by partially substituting some of the cobalt for tin and binding the material to a commercially available graphene film — a single-atom thick material that supports electronic materials without changing their properties — they could fabricate what they called a low-cost, easy-to-develop electrode.
Once the simulations were completed, the team in China conducted experiments to see if the simulation could be actualized.
“The experimental results verified a significantly increased conductivity of the cobalt oxide structure after partial substitution by tin,” Zhu said. “The developed device is expected to have promising practical applications as the next-generation energy storage device.”
Next, Zhu and Cheng plan to use their own version of graphene film — a porous foam created by partially cutting and then breaking the material with lasers — to fabricate a flexible capacitor to allow for easy and fast conductivity.
“The supercapacitor is one key component, but we’re also interested in combining with other mechanisms to serve as both an energy harvester and a sensor,” Cheng said. “Our goal is to put a lot of functions into a simple, self-powered device.”
The National Natural Science Foundation of China supported this work.
Story Source:
Materials provided by Penn State. Original written by Ashley WennersHerron. Note: Content may be edited for style and length. More