Computers Math

A novel artificial intelligence blood testing technology developed by researchers at the Johns Hopkins Kimmel Cancer Center was found to detect over 90% of lung cancers in samples from nearly 800 individuals with and without cancer.
The test approach, called DELFI (DNA evaluation of fragments for early interception), spots unique patterns in the fragmentation of DNA shed from cancer cells circulating in the bloodstream. Applying this technology to blood samples taken from 796 individuals in Denmark, the Netherlands and the U.S., investigators found that the DELFI approach accurately distinguished between patients with and without lung cancer.
Combining the test with analysis of clinical risk factors, a protein biomarker, and followed by computed tomography imaging, DELFI helped detect 94% of patients with cancer across stages and subtypes. This included 91% of patients with earlier or less invasive stage I/II cancers and 96% of patients with more advanced stage III/IV cancers. These results will be published in the August 20 issue of the journal Nature Communications.
Lung cancer is the most common cause of cancer death, claiming almost 2 million lives worldwide each year. However, fewer than 6% of Americans at risk for lung cancers undergo recommended low-dose computed tomography screening, despite projections that tens of thousands of deaths could be avoided, and even fewer are screened worldwide, explains senior study author Victor E. Velculescu, M.D., Ph.D., professor of oncology and do-director of the Cancer Genetics and Epigenetics Program at the Johns Hopkins Kimmel Cancer Center. This is due to a variety of reasons, including concerns of potential harm from investigation of false positive imaging results, radiation exposure or worries about complications from invasive procedures. “It is clear that there is an urgent, unmet clinical need for development of alternative, noninvasive approaches to improve cancer screening for high-risk individuals and, ultimately, the general population,” says lead author Dimitrios Mathios, a postdoctoral fellow at the Johns Hopkins Kimmel Cancer Center. “We believe that a blood test, or ‘liquid biopsy,’ for lung cancer could be a good way to enhance screening efforts, because it would be easy to do, broadly accessible and cost-effective.”
The DELFI technology uses a blood test to indirectly measure the way DNA is packaged inside the nucleus of a cell by studying the size and amount of cell-free DNA present in the circulation from different regions across the genome. Healthy cells package DNA like a well-organized suitcase, in which different regions of the genome are placed carefully in various compartments. The nuclei of cancer cells, by contrast, are like more disorganized suitcases, with items from across the genome thrown in haphazardly. When cancer cells die, they release DNA in a chaotic manner into the bloodstream. DELFI helps identify the presence of cancer using machine learning, a type of artificial intelligence, to examine millions of cell-free DNA fragments for abnormal patterns, including the size and amount of DNA in different genomic regions. This approach provides a view of cell-free DNA referred to as the “fragmentome.” The DELFI approach only requires low-coverage sequencing of the genome, enabling this technology to be cost-effective in a screening setting, the researchers say.
For the study, investigators from Johns Hopkins, working with researchers in Denmark and the Netherlands, first performed genome sequencing of cell-free DNA in blood samples from 365 individuals participating in a seven-year Danish study called LUCAS. The majority of participants were at high risk for lung cancer and had smoking-related symptoms such as cough or difficulty breathing. The DELFI approach found that patients who were later determined to have cancer had widespread variation in their fragmentome profiles, while patients found not to have cancer had consistent fragmentome profiles. Subsequently, researchers validated the DELFI technology using a different population of 385 individuals without cancer and 46 individuals with cancer. Overall, the approach detected over 90% of patients with lung cancer, including those with early and advanced stages, and with different subtypes. “DNA fragmentation patterns provide a remarkable fingerprint for early detection of cancer that we believe could be the basis of a widely available liquid biopsy test for patients with lung cancer,” says author Rob Scharpf, Ph.D., associate professor of oncology at the Johns Hopkins Kimmel Cancer Center.
A first-of-a-kind national clinical trial called DELFI-L101, sponsored by the Johns Hopkins University spin-out Delfi Diagnostics, is evaluating a test based on the DELFI technology in 1,700 participants in the U.S., including healthy participants, individuals with lung cancers and individuals with other cancers. The group would like to further study DELFI in other types of cancers.
Other scientists who contributed to the work include Stephen Cristiano, Jamie E. Medina, Jillian Phallen, Daniel Bruhm, Noushin Niknafs, Leonardo Ferreira, Vilmos Adleff, Jia Yuee Ciao, Alessandro Leal, Michael Noe, James White, Adith S. Arun, Carolyn Hruban, Akshaya V. Annapragada, Patrick M. Forde, Valsamo Anagnostou and Julie R. Brahmer of Johns Hopkins. Additional authors were from Herlev and Gentofte Hospital and Bispebjerg Hospital in Copenhagen; Aarhus University Hospital in Aarhus, Denmark; Herning Regional Hospital in Herning, Denmark; the Netherlands Cancer Institute in Amsterdam; Delfi Diagnostics; and Hvidovre Hospital in Hvidovre, Denmark.
The work was supported in part by the Dr. Miriam and Sheldon G. Adelson Medical Research Foundation; a Stand Up to Cancer /INTIME Lung Cancer Interception Dream Team grant; Stand Up to Cancer-Dutch Cancer Society International Translational Cancer Research Dream Team Grant (SU2C-AACR-DT1415); the Gray Foundation; the Commonwealth Foundation; the Mark Foundation for Cancer Research; the Lundbeck Foundation; an unrestricted grant from Roche Denmark; a research grant from Delfi Diagnostics; and National Institutes of Health grants CA121113, CA006973, CA233259 and 1T32GM136577.
Mathios, Cristiano, Phallen, Leal, Adleff, Scharpf and Velculescu are inventors on patent applications submitted by Johns Hopkins University related to cell-free DNA for cancer detection. Cristiano, Phallen, Leal, Adleff and Scharpf are founders of Delfi Diagnostics, and Adleff and Scharpf are consultants for this organization. Velculescu is a founder of Delfi Diagnostics and of Personal Genome Diagnostics, serves on the board of directors and as a consultant for both organizations, and owns Delfi Diagnostics and Personal Genome Diagnostics stock, which are subject to certain restrictions under university policy. The Johns Hopkins University owns equity in Delfi Diagnostics and Personal Genome Diagnostics. Additionally, Velculescu is an adviser to Bristol-Myers Squibb, Genentech, and Takeda Pharmaceuticals. The terms of these arrangements are managed by The Johns Hopkins University in accordance with its conflict of interest policies. More

One of the biggest marketing trends in the online shopping industry is personalization through curated product recommendations; however, it can change whether people buy a product they had been considering, according to new University of California San Diego research.
The study by Uma R. Karmarkar, assistant professor at the UC San Diego Rady School of Management and School of Global Policy and Strategy, finds that display items that come from the same category as the target product, such as a board game matched with other board games, enhance the chances of a target product’s purchase. In contrast, consumers are less likely to buy the target product if it is mismatched with products from different categories, for example, a board game displayed with kitchen knives.
The study utilized eye-tracking — a sensor technology that makes it possible to know where a person is looking — to examine how different types of displays influenced visual attention. Participants in the study looked at their target product for the same amount of time when it was paired with similar items or with items from different categories; however, shoppers spent more time looking at the mismatched products, even though they were only supposed to be there “for display.”
“What is surprising is that when I asked people how much they liked the target products, their preferences didn’t change between display settings,” Karmarkar said. “The findings show that it is not about how much you like or dislike the item you’re looking at, it’s about your process for buying the item. The surrounding display items don’t seem to change how much attention you give the target product, but they can influence your decision whether to buy it or not.”
Karmarkar, who holds PhDs in consumer behavior and neuroscience, says the findings suggests that seeing similar options on the page reinforces the idea to consumers that they’re making the right kind of decision to purchase an item that fits the category on display.
“When the information is mismatched, it changes the scope of the decision,” she said. “A mismatched display is comparable to shopping in a store with more variety. You may consider a featured board game but if you can see other products to buy, this board game may not be the first kind of purchase you want to make. The mismatched items draw additional attention and compete with the category you were considering.”
The study, to be published in Frontiers in Neuroscience, involved 58 participants, ages 18 to 40, who had to make 36 online shopping decisions for real products with real money. The findings showing differences in purchase rates replicate a set of studies Karmarkar published in 2017. In the new research, she was able to measure what parts of the display were engaging more or less attention. In addition, the upcoming paper shows that matched displays increase purchase rates even when they include more attention-grabbing information, like details about price.
Karmarkar talked with industry experts about product recommendations systems, which shaped her approach to these questions. Recommender algorithms can have different designs to meet a variety of retailers’ respective goals. Products can be shown with “mismatched” displays when retailers are using cross-promoting tactics based on prior customer behavior or inventory they may want to sell more rapidly.
The board game example Karmarkar often uses is based on a real experience she had while shopping online during the month of October.
“I had been browsing games like ‘Bananagrams’ and when I reloaded the product page, a Halloween costumes display popped up,” she said. “Given my search history, the store probably estimated I had a family. So while I’m sure they wanted me to buy the game, they also knew they had an active shopper who might be interested in the Halloween costumes that needed to sell by the end of the month. It looks like a win-win, but our work suggests that creating this mismatched situation could have lowered the chance that I would add the game to my cart.”
While the study is useful for online retailers to know the benefits of showing same-category options on a specific product page, the research is valuable to consumers as well.
“This shows how outside forces shape our decisions in ways we might not recognize,” she said. “If a shopper is looking for something specific, they are likely to focus their attention, regardless of recommender displays. But when people are just ‘browsing stuff online,’ different page designs can create different patterns of attention. Store displays can change what we choose, even when they don’t change what we like.” More

Since the beginning of the pandemic, researchers worldwide have been looking for ways to treat COVID-19. And while the COVID-19 vaccines represent the best measure to prevent the disease, therapies for those who do get infected remain in short supply. A new groundbreaking study from U-M reveals several drug contenders already in use for other purposes — including one dietary supplement — that have been shown to block or reduce SARS-CoV2 infection in cells.
The study, published recently in the Proceedings of the National Academy of Science, uses artificial intelligence-powered image analysis of human cell lines during infection with the novel coronavirus. The cells were treated with more than 1,400 individual FDA-approved drugs and compounds, either before or after viral infection, and screened, resulting in 17 potential hits. Ten of those hits were newly recognized, with seven identified in previous drug repurposing studies, including remdesivir, which is one of the few FDA-approved therapies for COVID-17 in hospitalized patients.
“Traditionally, the drug development process takes a decade — and we just don’t have a decade,” said Jonathan Sexton, Ph.D., Assistant Professor of Internal Medicine at the U-M Medical School and one of the senior authors on the paper. “The therapies we discovered are well positioned for phase 2 clinical trials because their safety has already been established.”
The team validated the 17 candidate compounds in several types of cells, including stem-cell derived human lung cells in an effort to mimic SARS-CoV2 infection of the respiratory tract. Nine showed anti-viral activity at reasonable doses, including lactoferrin, a protein found in human breastmilk that is also available over the counter as a dietary supplement derived from cow’s milk.
“We found lactoferrin had remarkable efficacy for preventing infection, working better than anything else we observed,” Sexton said. He adds that early data suggest this efficacy extends even to newer variants of SARS-CoV2, including the highly transmissible Delta variant.
The team is soon launching clinical trials of the compound to examine its ability to reduce viral loads and inflammation in patients with SARS-CoV2 infection. More

A study led by researchers at North Carolina State University developed a new method that enables quantum computers to measure the thermodynamic properties of systems by calculating the zeros of the partition function.
“We’ve illustrated a new way to get at thermodynamic properties of a system, such as free energy, entropy, and other properties that are too complex to currently be measured via traditional or quantum computing,” says Lex Kemper, associate professor of physics at NC State and corresponding author of a paper describing the work. “By calculating partition function zeros we are on the way to solving the problem of scaling to larger numbers of qubits when trying to calculate free energies and entropies in a given system.”
Quantum computers are often used to study complicated systems due to their ability to handle large computations beyond the reach of conventional computers. However, some problems, such as measuring the thermodynamics or free energy in a system (which involves calculating its entropy), are still too big for even these computers to handle efficiently.
A partition function describes the statistical properties of a system in thermodynamic equilibrium. The total energy, free energy, entropy, or pressure of a system can be expressed mathematically in terms of the partition function or its derivatives.
Kemper and his colleagues used a quantum computer to measure the partition function zeros, rather than the entropy, of a spin model as it is tuned across a phase transition.
“Our method skips the part where we calculate the entropy in favor of looking at the partition function,” Kemper says. “That’s because the partition function is a generating function — a function that you can perform operations on to get at other thermodynamic information such as the internal energy and the entropy.
“We measure the partition function by determining where it is zero. Once you know all the zeros of a function, you know the whole function. Since the zeros lie in the complex plane, we used a mapping between having a complex magnetic field and time evolution to find them.”
The researchers calculated the partition function on both a standard and a trapped ion quantum computer in the laboratory of Norbert Linke at the University of Maryland. The results from both compared favorably.
“This is a way to use a quantum computer to get at all the thermodynamic properties of a system without necessitating huge numbers of quantum computations,” Kemper says.
The research appears in Science Advances and is supported by the Department of Energy (grant DE-SC0019469). First author of the paper Akhil Francis is a graduate student at NC State. Norbert Linke and Chris Monroe from the University of Maryland; Jim Freericks from Georgetown University; and Sonika Johri from IonQ also contributed to the work.
Story Source:
Materials provided by North Carolina State University. Original written by Tracey Peake. Note: Content may be edited for style and length. More

Since the early days of the COVID pandemic, scientists have aggressively pursued the secrets of the mechanisms that allow severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) to enter and infect healthy human cells.
Early in the pandemic, University of California San Diego’s Rommie Amaro, a computational biophysical chemist, helped develop a detailed visualization of the SARS-CoV-2 spike protein that efficiently latches onto our cell receptors.
Now, Amaro and her research colleagues from UC San Diego, University of Pittsburgh, University of Texas at Austin, Columbia University and University of Wisconsin-Milwaukee have discovered how glycans — molecules that make up a sugary residue around the edges of the spike protein — act as infection gateways.
Published August 19 in the journal Nature Chemistry, a research study led by Amaro, co-senior author Lillian Chong at the University of Pittsburgh, first author and UC San Diego graduate student Terra Sztain and co-first author and UC San Diego postdoctoral scholar Surl-Hee Ahn, describes the discovery of glycan “gates” that open to allow SARS-CoV-2 entry.
“We essentially figured out how the spike actually opens and infects,” said Amaro, a professor of chemistry and biochemistry and a senior author of the new study. “We’ve unlocked an important secret of the spike in how it infects cells. Without this gate the virus basically is rendered incapable of infection.”
Amaro believes the research team’s gate discovery opens potential avenues for new therapeutics to counter SARS-CoV-2 infection. If glycan gates could be pharmacologically locked in the closed position, then the virus is effectively prevented from opening to entry and infection. More

‘Growing’ electronic components directly onto a semiconductor block avoids messy, noisy oxidation scattering that slows and impedes electronic operation.
A UNSW study out this month shows that the resulting high-mobility components are ideal candidates for high-frequency, ultra-small electronic devices, quantum dots, and for qubit applications in quantum computing.
Smaller Means Faster, but Also Noisier
Making computers faster requires ever-smaller transistors, with these electronic components now only a handful of nanometres in size. (There are around 12 billion transistors in the postage-stamp sized central chip of modern smartphones.)
However, in even smaller devices, the channel that the electrons flow through has to be very close to the interface between the semiconductor and the metallic gate used to turn the transistor on and off. Unavoidable surface oxidation and other surface contaminants cause unwanted scattering of electrons flowing through the channel, and also lead to instabilities and noise that are particularly problematic for quantum devices.
“In the new work we create transistors in which an ultra-thin metal gate is grown as part of the semiconductor crystal, preventing problems associated with oxidation of the semiconductor surface,” says lead author Yonatan Ashlea Alava. More

Scientists from SANKEN at Osaka University demonstrated the readout of spin-polarized multielectron states composed of three or four electrons on a semiconductor quantum dot. By making use of the spin filtering caused by the quantum Hall effect, the researchers were able to improve upon previous methods that could only easily resolve two electrons. This work may lead to quantum computers based on the multielectron high-spin states.
Despite the almost unimaginable increase in the power of computers over the last 75 years, even the fastest machines available today run on the same basic principle as the original room-sized collection of vacuum tubes: information is still processed by herding electrons through circuits based on their electric charge. However, computer manufacturers are rapidly reaching the limit of how much they can readily achieve with charge alone, and new methods, such as quantum computing, are not ready yet to take their place. One promising approach is to utilize the intrinsic magnetic moment of electrons, called “spin,” but controlling and measuring these values has proven to be very challenging.
Now, a team of researchers led by Osaka University showed how to read out the spin state of multiple electrons confined to a tiny quantum dot fabricated from gallium and arsenic. Quantum dots act like artificial atoms with properties that can be tuned by scientists by changing their size or composition. However, the gaps in energy levels generally becomes smaller and harder to resolve as the number of trapped electrons increases.
To overcome this, the team took advantage of a phenomenon called the quantum Hall effect. When electrons are confined to two dimensions and subjected to a strong magnetic field, their states become quantized, so their energy levels can only take on certain specific values. “Previous spin readout methods could only handle one or two electrons, but using the quantum Hall effect, we were able to resolve up to four spin-polarized electrons,” first author Haruki Kiyama says. To prevent disturbances from thermal fluctuations, the experiments were performed at extremely low temperatures, around 80 millikelvin. “This readout technique may pave the way toward faster and higher-capacity spin-based quantum information processing devices with multielectron spin states,” senior author Akira Oiwa says.
Story Source:
Materials provided by Osaka University. Note: Content may be edited for style and length. More

Whether a robot or a person delivers your package, the carbon footprint would essentially be the same, according to a University of Michigan study that could help inform the future of automated delivery as the pandemic fuels a dramatic rise in online shopping.
The researchers examined the environmental impacts of advanced residential package delivery scenarios that use electric and gas-powered autonomous vehicles and two-legged robots to ferry goods from delivery hubs to neighborhoods, and then to front doors. They compared those impacts with the traditional approach of a human driver who hand-delivers parcels.
They found that while robots and automation contribute less than 20% of a package’s footprint, most of the greenhouse gas emissions come from the vehicle. Vehicle powertrain and fuel economy are the key factors determining the package’s footprint. Switching to electric vehicles and reducing the carbon intensity of the electricity they run on could have the biggest impacts in sustainable parcel delivery, the researchers say.
Their study is a life cycle analysis of the cradle-to-grave greenhouse gas emissions for 12 suburban delivery scenarios. It’s unique in that it doesn’t just tally emissions from the delivery process. It also counts greenhouse gases from manufacturing the vehicles and robots, as well as disposing of them or recycling them at the end of their lives.
“We found that the energy and carbon footprints of this automated parcel delivery in suburban areas was similar to that of conventional human driven vehicles. The advantages of better fuel economy through vehicle automation were offset by greater electricity loads from automated vehicle power requirements,” said Gregory Keoleian, the Peter M. Wege Endowed Professor of Sustainable Systems at the U-M School for Environment and Sustainability and a professor of civil and environmental engineering.
“For all delivery systems studied, the vehicle-use phase is the single largest contributor to greenhouse gas emissions, highlighting the need for low-carbon fuels for sustainable parcel delivery. It is critically important to decarbonize grids while deploying electrified vehicles.”
Optimizing ‘the last mile’ in a surging package delivery market More