Posted on

Terahertz wireless could lead to fiber-optics speed in-flight and mobile metropolitan internet

Terahertz wireless links to spaceborne satellites could one day make gigabit-per-second connection speeds available to anyone, anytime, anywhere on the face of the earth, on the ground or in flight (credit: Fujishima et al./Hiroshima University)

Hiroshima University researchers and associates have developed a terahertz* (THz) transmitter capable of transmitting digital data over a single channel at a speed of 105 gigabits per second (Gbps), and demonstrated the technology at the International Solid-State Circuits Conference (ISSCC) 2017 conference last week.

For perspective, that’s more than 100 times faster than the fastest (1 Gbps) internet connection in the U.S. or more than 3,000 times faster than the 31 Mbps available to the average U.S. household in 2014, according to an FCC report. It’s also ten times or more faster than the fastest rate expected to be offered by fifth-generation mobile networks (5G) for metropolitan areas around 2020.

Major uses: faster in-flight and metropolitan internet, high-frequency trading

Applications of this forthcoming THz technology include higher-speed in-flight network connection speeds via satellite, fast download of videos and other large files for mobile devices, and ultrafast wireless links between base stations, according to Hiroshima University professor Minoru Fujishima.

An important business application is faster high-frequency trading, which requires minimal latency (delay). Recently, the time it takes to execute these trades has gone from milliseconds (thousandths of a second) to microseconds (millionths of a second), as KurzweilAI has explained. However, long-distance fiber optics connections (currently used for long-distance trading) have significant latency because light (as radio waves) travels 50%  faster in a vacuum than through glass fiber, while microwaves traveling in air have a less than a 1% speed reduction.

The National Institute of Information and Communications Technology and Panasonic Corporation are also partners in this research.

* Terahertz, a frequency range that is 1,000 times higher than gigahertz, or 1012 Hz, actually starts at 100 GHz or .1 THz. The researchers transmitted in this unregulated THz band — a vast new frequency resource expected to be used for future ultrahigh-speed wireless communications — using the frequency range from 290 GHz to 315 GHz. The full range of frequencies in the THz band (275 GHz to 450 GHz) are currently unallocated, but are expected to be discussed at the World Radiocommunication Conference 2019.

Abstract of A 105Gb/s 300GHz CMOS Transmitter

“High speed” in communications often means “high data-rate” and fiber-optic technologies have long been ahead of wireless technologies in that regard. However, an often overlooked definite advantage of wireless links over fiber-optic links is that waves travel at the speed of light c, which is about 50% faster than in optical fibers as shown in Fig. 17.9.1 (top left). This “minimum latency” is crucial for applications requiring real-time responses over a long distance, including high-frequency trading [1]. Further opportunities and new applications might be created if the absolute minimum latency and fiber-optic data-rates are put together. (Sub-)THz frequencies have an extremely broad atmospheric transmission window with manageable losses as shown in Fig. 17.9.1 (top right) and will be ideal for building light-speed links supporting fiber-optic data-rates. This paper presents a 105Gb/s 300GHz transmitter (TX) fabricated using a 40nm CMOS process.

Abstract of A 300GHz 40nm CMOS transmitter with 32-QAM 17.5Gb/s/ch capability over 6 channels

The vast unallocated frequency band lying above 275GHz offers enormous potential for ultrahigh-speed wireless communication. An overall bandwidth that could be allocated for multi-channel communication can easily be several times the 60GHz unlicensed bandwidth of 9GHz. We present a 300GHz transmitter (TX) in 40nm CMOS, capable of 32-quadrature amplitude modulation (QAM) 17.5Gb/s/ch signal transmission. It can cover the frequency range from 275 to 305GHz with 6 channels as shown at the top of Fig. 20.1.1. Figure 20.1.1 also lists possible THz TX architectures, based on recently reported above-200GHz TXs. The choice of architecture depends very much on the transistor unity-power-gain frequency fmax. If the fmax is sufficiently higher than the carrier frequency, the ordinary power amplifier (PA)-last architecture (Fig. 20.1.1, top row of the table) is possible and preferable [1-3], although the presence of a PA is, of course, not a requirement [4,5]. If, on the other hand, the fmax is comparable to or lower than the carrier frequency as in our case, a PA-less architecture must be adopted. A typical such architecture is the frequency multiplier-last architecture (Fig. 20.1.1, middle row of the table). For example, a 260GHz quadrupler-last on-off keying (OOK) TX [6] and a 434GHz tripler-last amplitude-shift keying (ASK) TX [7] were reported. A drawback of this architecture is the inefficient bandwidth utilization due to signal bandwidth spreading. Another drawback is that the use of multibit digital modulation is very difficult, if not impossible. An exception to this is the combination of quadrature phase-shift keying (QPSK) and frequency tripling. When a QPSK-modulated intermediate frequency (IF) signal undergoes frequency tripling, the resulting signal constellation remains that of QPSK with some symbol permutation. Such a tripler-last 240GHz QPSK TX was reported [8]. However, a 16-QAM constellation, for example, would suffer severe distortion by frequency tripling. If the 300GHz band is to be seriously considered for a platform for ultrahigh-speed wireless communication, QAM-capability will be a requisite.

Posted on

New machine-learning algorithms may revolutionize drug discovery — and our understanding of life

A new set of machine-learning algorithms can generate 3D structures of complex nanoscale protein molecules like this complex proteasome map refined to 2.8 Angstroms (.28 nanometer) in 70 min with 49,954 particle images (credit: Structura Biotechnology Inc.)

A new set of machine-learning algorithms developed by researchers at the University of Toronto Scarborough can generate 3D structures of nanoscale protein molecules that could not be achieved in the past. The algorithms may revolutionize the development of new drug therapies for a range of diseases and may even lead to better understand how life works at the atomic level, the researchers say.

Drugs work by binding to a specific protein molecule and changing the protein’s 3D shape, which alters the way the drug works once inside the body. The ideal drug is designed in a shape that will only bind to a specific protein or group of proteins that are involved in a disease, while eliminating side effects that occur when drugs bind to other proteins in the body.

A significant computational problem

Since proteins are tiny — about 1 to 100 nanometers — even smaller than the shortest wavelength of visible light, they can’t be seen directly without using sophisticated techniques like electron cryomicroscopy (cryo-EM). Cryo-EM uses high-power microscopes to take tens of thousands of low-resolution images of a frozen protein sample from different positions.

The computational problem is to then piece together the correct high-resolution 3D structure from these 2D images.

Existing techniques take several days or even weeks to generate a 3D structure on a cluster of computers, requiring as much as 500,000 CPU hours, according to the researchers. Also, existing techniques often generate incorrect structures unless an expert user provides an accurate guess of the molecule being studied.

CryoSPARC machine learning algorithms can generate 3-D structures of nanoscale protein molecules (credit: Structura Biotechnology Inc)

New high-speed, deep-learning algorithms

That’s where the new set of algorithms* comes in. It reconstructs 3D structures of protein molecules using these images. “Our approach solves some of the major problems in terms of speed and number of structures you can determine,” says Professor David Fleet, chair of the Computer and Mathematical Sciences Department at U of Toronto Scarborough.

The algorithms could significantly aid in the development of new drugs because they provide a faster, more efficient means at arriving at the correct protein structure.

The new approach, called cryoSPARC, developed by the team’s startup, Structura Biotechnology Inc., eliminates the need for that prior knowledge and can make the computations possible in minutes on a single computer, using a standalone graphics processing unit (GPU) accelerated software package, according to the researchers.

The research was published in the current edition of the journal Nature Methods. It received funding from the Natural Sciences and Engineering Research Council of Canada (NSERC). The new cryo-EM platform is already being used in labs across North America, the researchers note.

* “We use an SGD [stochastic gradient descent] optimization scheme to quickly identify one or several low-resolution 3D structures that are consistent with a set of observed images. This algorithm allows for ab initio heterogeneous structure determination with no prior model of the molecule’s structure. Once approximate structures are determined, a branch-and-bound algorithm for image alignment helps rapidly refine structures to high resolution. The speed and robustness of these approaches allow structure determination in a matter of minutes or hours on a single inexpensive desktop workstation. … SGD was popularized as a key tool in deep learning for the optimization of nonconvex functions, and it results in near human-level performance in tasks like image and speech recognition.” — Ali Punjani et al./Nature Methods

University of Toronto Scarborough | New algorithms may revolutionize drug discoveries and our understanding of life

Abstract of cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination

Single-particle electron cryomicroscopy (cryo-EM) is a powerful method for determining the structures of biological macromolecules. With automated microscopes, cryo-EM data can often be obtained in a few days. However, processing cryo-EM image data to reveal heterogeneity in the protein structure and to refine 3D maps to high resolution frequently becomes a severe bottleneck, requiring expert intervention, prior structural knowledge, and weeks of calculations on expensive computer clusters. Here we show that stochastic gradient descent (SGD) and branch-and-bound maximum likelihood optimization algorithms permit the major steps in cryo-EM structure determination to be performed in hours or minutes on an inexpensive desktop computer. Furthermore, SGD with Bayesian marginalization allows ab initio 3D classification, enabling automated analysis and discovery of unexpected structures without bias from a reference map. These algorithms are combined in a user-friendly computer program named cryoSPARC

Posted on

Scientists create first 3-D synchronized-beating heart tissue

3-D fluorescent image of 3-D tissue with many cells laid down sequentially to create attached layers of alternating cell types, like membranes in the human body (credit: York University)

York University scientists have created the first in vitro (lab) 3D heart tissue made from three different types of cardiac cells that beat in synchronized harmony. It may lead to better understanding of cardiac health and improved treatments.*

The researchers constructed the heart tissue from three free-beating rat cell types: contractile cardiac muscle cells, connective tissue cells, and vascular cells. No external scaffold was used and the cells were the only building blocks of the generated cardiac tissue. The researchers believe this is the first 3D in vitro cardiac tissue with three cell types that can beat together as one entity, rather than at different intervals, with high cell density and efficient cell contacts, and without the requirement of external electrical stimulation.

The substance used to stick the cells together (ViaGlue) may also provide researchers with tools to create and test 3D in vitro cardiac tissue in their own labs to study heart disease and issues with transplantation.

“This breakthrough will allow better and earlier drug testing, and potentially eliminate harmful or toxic medications sooner,” said York U chemistry Professor Muhammad Yousaf.

For 2D or 3D cardiac tissue to be functional it needs the same high cellular density and the cells must be in contact to facilitate synchronized beating, according to the researchers. The 3D cardiac tissue was created at a millimeter scale, but larger versions could be made, said Yousaf, who has created a start-up company, OrganoLinX, to commercialize the ViaGlue reagent and to provide custom 3D tissues on demand.

“Production of 3-dimensional artificial cardiac tissues for fundamental studies of heart disease, transplantation, and evaluation of drug toxicity is an important and intense area of research,” the researchers note in a paper in open-access Nature Scientific Reports.

* Cardiovascular-associated diseases are the leading cause of death globally and are responsible for 40 per cent of deaths in North America, according to a 2011 report from the American Heart Association.


York University | York U makes a 3D heart beat as one

Abstract of Scaffold Free Bio-orthogonal Assembly of 3-Dimensional Cardiac Tissue via Cell Surface Engineering

There has been tremendous interest in constructing in vitro cardiac tissue for a range of fundamental studies of cardiac development and disease and as a commercial system to evaluate therapeutic drug discovery prioritization and toxicity. Although there has been progress towards studying 2-dimensional cardiac function in vitro, there remain challenging obstacles to generate rapid and efficient scaffold-free 3-dimensional multiple cell type co-culture cardiac tissue models. Herein, we develop a programmed rapid self-assembly strategy to induce specific and stable cell-cell contacts among multiple cell types found in heart tissue to generate 3D tissues through cell-surface engineering based on liposome delivery and fusion to display bio-orthogonal functional groups from cell membranes. We generate, for the first time, a scaffold free and stable self assembled 3 cell line co-culture 3D cardiac tissue model by assembling cardiomyocytes, endothelial cells and cardiac fibroblast cells via a rapid inter-cell click ligation process. We compare and analyze the function of the 3D cardiac tissue chips with 2D co-culture monolayers by assessing cardiac specific markers, electromechanical cell coupling, beating rates and evaluating drug toxicity.

Posted on

First stable semisynthetic organism created

DNA — now with a new base pair! (credit: Romesberg Lab)

Scientists at The Scripps Research Institute (TSRI) have developed the first stable semisynthetic organism — a bacterium with two new synthetic bases (called X and Y) added to the four natural bases (A, T, C, and G) that every living organism possesses. Adding two more letters to expand the genetic alphabet can be used to make novel proteins for new therapeutics, according to the researchers.

All life as we currently know it contains just four bases that pair up to form two “base pairs” — the rungs of the DNA ladder — which are simply rearranged to create different organisms.

DNA double helix with examples of base pairs (credit: National Human Genome Research Institute)

Until now. TSRI Professor Floyd Romesberg and his colleagues have now shown that their new “more life-like” single-celled organism can hold on indefinitely to the new synthetic base pair as it divides. Their research was published January 23, 2017, online ahead of print in the journal Proceedings of the National Academy of Sciences.

Snipping together a new semisynthetic organism

Top: chemical structure of a normal DNA base pair, with C and G bases. Below: structures of the two new X and Y synthetic bases. (credit: Yorke Zhang et al./PNAS)

To ensure the new semisynthetic organism retains the new synthetic base pair as it divides, the researchers used CRISPR-Cas9 (a sort of scissors for modifying DNA). They engineered the organism to react to a genetic sequence that doesn’t have X and Y as a foreign invader (an immune response).

So any new cell that dropped X and Y would be marked for destruction — leaving the scientists with an organism that could hold on to the new bases.

In their experiments, that enabled their semisynthetic organism to keep X and Y in its genome after dividing 60 times, leading the researchers to believe it can hold on to the new base pair indefinitely.

“We can now get the light of life to stay on,” said Romesberg, senior author of the new study. “That suggests that all of life’s processes can be subject to manipulation.”

Romesberg emphasized that this work is only in single cells and is not meant to be used in more complex organisms. So far, scientists can only get the organism to store genetic information.

While applications for this kind of organism are still far in the future, the researchers say the work could be used to create new functions for single-celled organisms that play important roles in drug discovery and “much more.” (Romesberg has created created a biotech company named Synthorx for discovering and developing novel protein therapeutics.)

“All of life’s processes can be subject to manipulation”

Next, the researchers plan to study how their new genetic code can be transcribed into RNA, the molecule in cells needed to translate DNA into useful proteins.

Scientists at the University of Grenoble and Henan Normal University were also involved in the study, which was supported by grants from the National Institutes of Health, theNational Science Foundation, the National Natural Science Foundation of China, Labex ARCANE, NanoBio-ICMG platforms, and a postdoctoral fellowship from the American Cancer Society.

Abstract of A semisynthetic organism engineered for the stable expansion of the genetic alphabet

All natural organisms store genetic information in a four-letter, two-base-pair genetic alphabet. The expansion of the genetic alphabet with two synthetic unnatural nucleotides that selectively pair to form an unnatural base pair (UBP) would increase the information storage potential of DNA, and semisynthetic organisms (SSOs) that stably harbor this expanded alphabet would thereby have the potential to store and retrieve increased information. Toward this goal, we previously reported that Escherichia coli grown in the presence of the unnatural nucleoside triphosphates dNaMTP and d5SICSTP, and provided with the means to import them via expression of a plasmid-borne nucleoside triphosphate transporter, replicates DNA containing a single dNaM-d5SICS UBP. Although this represented an important proof-of-concept, the nascent SSO grew poorly and, more problematically, required growth under controlled conditions and even then was unable to indefinitely store the unnatural information, which is clearly a prerequisite for true semisynthetic life. Here, to fortify and vivify the nascent SSO, we engineered the transporter, used a more chemically optimized UBP, and harnessed the power of the bacterial immune response by using Cas9 to eliminate DNA that had lost the UBP. The optimized SSO grows robustly, constitutively imports the unnatural triphosphates, and is able to indefinitely retain multiple UBPs in virtually any sequence context. This SSO is thus a form of life that can stably store genetic information using a six-letter, three-base-pair alphabet.

Posted on

Beneficial AI conference develops ‘Asilomar AI principles’ to guide future AI research

Beneficial AI conference (credit: Future of Life Institute)

At the Beneficial AI 2017 conference, January 5–8 held at a conference center in Asilomar, California — a sequel to the 2015 AI Safety conference in Puerto Rico — the Future of Life Institute (FLI) brought together more 100 AI researchers from academia and industry and thought leaders in economics, law, ethics, and philosophy to address and formulate principles of beneficial AI.

FLI hosted a two-day workshop for its grant recipients, followed by a 2.5-day conference, in which people from various AI-related fields hashed out opportunities and challenges related to the future of AI and steps we can take to ensure that the resulting technology is beneficial.

Beneficial AI conference participants (credit: Future of Life Institute)

The result was 23 Asilomar AI Principles, intended to suggest AI research guidelines, such as “The goal of AI research should be to create not undirected intelligence, but beneficial intelligence” and “An arms race in lethal autonomous weapons should be avoided”; identify ethics and values, such as safety and transparency; and address longer-term issues — notably, “ Superintelligence should only be developed in the service of widely shared ethical ideals, and for the benefit of all humanity rather than one state or organization.”

To date, 2515 AI researchers and others are signatories of the Principles. The process is described here.

The conference location has historic significance. In 2009, the Association for the Advancement of Artificial Intelligence held the Asilomar Meeting on Long-Term AI Futures to address similar concerns. And in 1975, the Asilomar Conference on Recombinant DNA was held to discuss potential biohazards and regulation of emerging biotechnology.

The non-profit Future of Life Institute was founded in March 2014 by MIT cosmologist Max Tegmark, Skype co-founder Jaan Tallinn, DeepMind research scientist Viktoriya Krakovna, Boston University Ph.D. candidate in Developmental Sciences Meia Chita-Tegmark, and UCSC physicist Anthony Aguirre. Its mission is “to catalyze and support research and initiatives for safeguarding life and developing optimistic visions of the future, including positive ways for humanity to steer its own course considering new technologies and challenges.”

FLI’s scientific advisory board includes physicist Stephen Hawking, SpaceX CEO Elon Musk, Astronomer Royal Martin Rees, and UC Berkeley Professor of Computer Science/Smith-Zadeh Professor in Engineering Stuart Russell.


Future of Life Institute | Superintelligence: Science or Fiction? | Elon Musk & Other Great Minds

Elon Musk, Stuart Russell, Ray Kurzweil, Demis Hassabis, Sam Harris, Nick Bostrom, David Chalmers, Bart Selman, and Jaan Tallinn discuss with Max Tegmark (moderator) what likely outcomes might be if we succeed in building human-level AGI [artificial general intelligence] (and beyond), and also what we would like to happen.

 

Posted on

Brain-computer interface enables completely locked-in patients to communicate for the first time

NIRS/EEG brain computer interface system applied to a model (credit: Wyss Center for Bio and Neuroengineering)

Four advanced ALS (amyotrophic lateral sclerosis) patients who were “completely locked in” (totally unable to communicate) for years have suddenly broken through in a lab at the Wyss Center for Bio and Neuroengineering in Geneva, Switzerland — communicating a “yes” or “no” by simply thinking the answers.

The brain–computer interface (BCI) system achieved this remarkable breakthrough by using functional near-infrared spectroscopy (fNIRS) to measure changes in blood oxygen levels in the frontal lobes of the brain.

Patients suffering from ALS paralysis, but with preserved awareness, cognition, eye movements and blinking, are classified as having “locked-in syndrome” and can communicate via a BCI by looking at a computer screen, for example. But when the disorder progresses until the patient loses control of the last muscular response, usually the eye muscles, the condition is known as “completely locked-in state” (CLIS), with no possibility of communication. For most of us: a nightmare world.

“Are you happy?”

But surprisingly, when the researchers asked the question, “Are you happy?,” the answer from all four was consistently “yes,” repeated over weeks of questioning. In response to the researchers’ statement, “I love to live,” three of the four replied yes. The researchers asked other personal questions that required “yes” or “no” answers, such as: “Your husband’s name is Joachim?”

They found the questions elicited correct responses in 70% of the trials. In one case, a family requested that the researchers asked the patient whether he would agree for his daughter to marry her boyfriend, Mario. The answer: “no” nine times out of ten.

Overturning previous theories, the research was published in an open-access paper January 31 in PLoS Biology. It was conducted by a multinational team led by Professor Niels Birbaumer, affiliated with the University of Tübingen in Germany; Ospedale San Camillo, IRCCS, Venice, Italy; and the Wyss Center for Bio and Neuroengineering. “If we could make this technique widely clinically available, it could have a huge impact on the day-to-day life of people with completely locked-in syndrome,” said Birbaumer.

“Restoring communication for completely locked-in patients is a crucial first step in the challenge to regain movement,” said Professor John Donoghue, Director of the Wyss Center. “The Wyss Center plans to build on the results of this study to develop clinically useful technology that will be available to people with paralysis resulting from ALS, stroke, or spinal cord injury. The technology used in the study also has broader applications that we believe could be further developed to treat and monitor people with a wide range of neuro-disorders.”

How fNIRS detected “yes” and “no”

One measurement on one channel of relative changes in oxygenated hemoglobin levels (vertical) vs. seconds (horizontal) for “yes” (left) and “no” (right) questions. (credit: Ujwal Chaudhary et al./PLos Biology)

The brain-computer interface in the study was based on functional near-infrared spectroscopy (fNIRS), which measures blood oxygenation (O2Hb). While other brain-computer interfaces have previously enabled some paralyzed patients to communicate, near-infrared spectroscopy is, so far, the only successful approach to restore communication to patients suffering from completely locked-in syndrome, according to the researchers.

After training a classifier separating “yes” from “no” answers for several days, the patients were given feedback of their affirmative or negative response to questions with known answers and open questions over a period of weeks. To measure relative change in oxygenated hemoglobin in the blood (indicating neural changes), the researchers used a NIRSport functional near-infrared spectroscopy system, which provides eight near-infrared sources and eight detectors, and placed these “optodes” over the frontocentral brain region.

To measure changes in O2Hb, the fNIRS system shined two wavelengths (760 nm and 850 nm) of pulsed near-infrared light. The blood component hemoglobin scatters light, and the ratio of infrared light absorbed to light scattered depends on the amount of hemoglobin binding with oxygen. NIRS measures the change of this ratio and infers the change in O2Hb concentration from that change. The NIRS device can reach to about 3 centimeters in the brain, with a resolution on the order of 5–10 mm, according to NIRX.

The work was supported by Deutsche Forschungsgemeinschaft; Stiftung Volkswagenwerk; German Ministry of Education and Research; Baden-Wuerttemberg Stiftung EMOIO from the Federal Ministry of Education and Research; Eva and Horst Koehler-Stiftung; National Natural Science Foundation of China; EU grant LUMINOUS; San Camillo hospital; and NINDS, NIH.


Brain-computer interface allows completely locked-in people to communicate

Posted on

Carnegie Mellon AI beats top poker pros — a first

“Brains vs Artificial Intelligence” competition at the Rivers Casino in Pittsburgh (credit: Carnegie Mellon University)

Libratus, an AI developed by Carnegie Mellon University, has defeated four of the world’s best professional poker players in a marathon 120,000 hands of Heads-up, No-Limit Texas Hold’em poker played over 20 days, CMU announced today (Jan. 31) — joining Deep Blue (for chess), Watson, and Alpha Go as major milestones in AI.

Libratus led the pros by a collective $1,766,250 in chips.* The tournament was held at the Rivers Casino in Pittsburgh from 11–30 January in a competition called “Brains Vs. Artificial Intelligence: Upping the Ante.”

The developers of Libratus — Tuomas Sandholm, professor of computer science, and Noam Brown, a Ph.D. student in computer science — said the sizable victory is statistically significant and not simply a matter of luck. “The best AI’s ability to do strategic reasoning with imperfect information has now surpassed that of the best humans,” Sandholm said. “This is the last frontier, at least in the foreseeable horizon, in game-solving in AI.”

This new AI milestone has implications for any realm in which information is incomplete and opponents sow misinformation, said Frank Pfenning, head of the Computer Science Department in CMU’s School of Computer Science. Business negotiation, military strategy, cybersecurity, and medical treatment planning could all benefit from automated decision-making using a Libratus-like AI.

“The computer can’t win at poker if it can’t bluff,” Pfenning explained. “Developing an AI that can do that successfully is a tremendous step forward scientifically and has numerous applications. Imagine that your smartphone will someday be able to negotiate the best price on a new car for you. That’s just the beginning.”

How the pros taught Libratus about its weaknesses

Brains vs AI scorecard (credit: Carnegie Mellon University)

So how was Libratus was able to improve day to day during the competition? It turns out it was the pros themselves who taught Libratus about its weaknesses. “After play ended each day, a meta-algorithm analyzed what holes the pros had identified and exploited in Libratus’ strategy,” Sandholm explained. “It then prioritized the holes and algorithmically patched the top three using the supercomputer each night.

“This is very different than how learning has been used in the past in poker. Typically researchers develop algorithms that try to exploit the opponent’s weaknesses. In contrast, here the daily improvement is about algorithmically fixing holes in our own strategy.”

Sandholm also said that Libratus’ end-game strategy was a major advance. “The end-game solver has a perfect analysis of the cards,” he said. It was able to update its strategy for each hand in a way that ensured any late changes would only improve the strategy. Over the course of the competition, the pros responded by making more aggressive moves early in the hand, no doubt to avoid playing in the deep waters of the endgame where the AI had an advantage, he added.

Converging high-performance computing and AI

Professor Tuomas Sandholm, Carnegie Mellon School of Computer Science, with the Pittsburgh Supercomputing Center’s Bridges supercomputer (credit: Carnegie Mellon University)

Libratus’ victory was made possible by the Pittsburgh Supercomputing Center’s Bridges computer. Libratus recruited the raw power of approximately 600 of Bridges’ 846 compute nodes. Bridges’ total speed is 1.35 petaflops, about 7,250 times as fast as a high-end laptop, and its memory is 274 terabytes, about 17,500 as much as you’d get in that laptop. This computing power gave Libratus the ability to play four of the best Texas Hold’em players in the world at once and beat them.

“We designed Bridges to converge high-performance computing and artificial intelligence,” said Nick Nystrom, PSC’s senior director of research and principal investigator for the National Science Foundation-funded Bridges system. “Libratus’ win is an important milestone toward developing AIs to address complex, real-world problems. At the same time, Bridges is powering new discoveries in the physical sciences, biology, social science, business and even the humanities.”

Sandholm said he will continue his research push on the core technologies involved in solving imperfect-information games and in applying these technologies to real-world problems. That includes his work with Optimized Markets, a company he founded to automate negotiations.

“CMU played a pivotal role in developing both computer chess, which eventually beat the human world champion, and Watson, the AI that beat top human Jeopardy! competitors,” Pfenning said. “It has been very exciting to watch the progress of poker-playing programs that have finally surpassed the best human players. Each one of these accomplishments represents a major milestone in our understanding of intelligence.

Head’s-Up No-Limit Texas Hold’em is a complex game, with 10160 (the number 1 followed by 160 zeroes) information sets — each set being characterized by the path of play in the hand as perceived by the player whose turn it is. The AI must make decisions without knowing all of the cards in play, while trying to sniff out bluffing by its opponent. As “no-limit” suggests, players may bet or raise any amount up to all of their chips.

Sandholm will be sharing Libratus’ secrets now that the competition is over, beginning with invited talks at the Association for the Advancement of Artificial Intelligence meeting Feb. 4–9 in San Francisco and in submissions to peer-reviewed scientific conferences and journals.

* The pros — Dong Kim, Jimmy Chou, Daniel McAulay and Jason Les — will split a $200,000 prize purse based on their respective performances during the event. McAulay, of Scotland, said Libratus was a tougher opponent than he expected, but it was exciting to play against it. “Whenever you play a top player at poker, you learn from it,” he said.


Carnegie Mellon University | Brains Vs. AI Rematch: Why Poker?

Posted on

Mayo Clinic researchers find mentally stimulating activities after age 70 associated with lower new cognitive-impairment risk

Mentally stimulating activities after age 70 found associated with decreased risk of new-onset mild cognitive impairment (credit: Mayo Clinic)

Mayo Clinic researchers have found that engaging in mentally stimulating activities, even after age 70, was associated with decreased risk of new-onset mild cognitive impairment (the intermediate stage between normal cognitive aging and dementia) over an average study period of 4 years.

The study discovered that for cognitively normal people 70 or older, the risk of new-onset mild cognitive impairment decreased by 30 percent with computer use, 28 percent with craft activities, 23 percent with social activities, and 22 percent with playing games — at least one to two times per week.*

The researchers found that persons who performed these activities had less cognitive decline than those who engaged in the same activities only two to three times per month or less,” says Yonas Geda, M.D., psychiatrist and behavioral neurologist at Mayo Clinic’s Arizona campus and senior author of the study.**

“Even for a person who is at genetic risk for cognitive decline***, engaging in some activities was beneficial,” says Janina Krell-Roesch, Ph.D., the first author of the study and a postdoctoral researcher in Dr. Geda’s Translational Neuroscience and Aging Program (TAP).

The results are published in an open-access article in the Jan. 30 edition of JAMA Neurology. Video.

* The researchers followed 1,929 cognitively normal participants of the population-based Mayo Clinic Study of Aging in Olmsted County, Minn., for an average duration of four years. The researchers conducted a neurocognitive assessment at the time of enrollment in the study, with evaluations every 15 months. They adjusted for sex, age and educational level. Following the assessment, an independent expert consensus panel at the Alzheimer Disease Research Center at Mayo Clinic made the classification of normal cognition or mild cognitive impairment for each study participant, based on published criteria. 

** The researchers note in the paper that a limitation of the study “pertains to potential recall bias that stems from the questionnaire on self-reported mentally stimulating activities. Also, we did not control for mentally stimulating activities performed in early life or mid-life. We can assume that individuals who engaged in mentally stimulating activities in early life or mid-life are more likely to engage in these activities in late life compared with persons who did not engage in these activities during the life span. Furthermore, an observational study like ours allows investigating associations but does not permit drawing conclusions about cause and effect, which can only be done by interventional (experimental) studies. Therefore, we cannot exclude a ‘reverse causality’ explanation (i.e., it is possible that participants who are at higher risk for MCI are less likely to engage in mentally stimulating activities).”

*** The benefits of being cognitively engaged were even seen among apolipoprotein E (APOE) ε4 carriers. APOE ε4 is a genetic risk factor for mild cognitive impairment and Alzheimer’s dementia. However, for APOE ε4 carriers, only computer use and social activities were associated with a decreased risk of mild cognitive impairment.

Abstract of Association Between Mentally Stimulating Activities in Late Life and the Outcome of Incident Mild Cognitive Impairment, With an Analysis of the APOE ε4 Genotype

Importance  Cross-sectional associations between engagement in mentally stimulating activities and decreased odds of having mild cognitive impairment (MCI) or Alzheimer disease have been reported. However, little is known about the longitudinal outcome of incident MCI as predicted by late-life (aged ≥70 years) mentally stimulating activities.

Objectives  To test the hypothesis of an association between mentally stimulating activities in late life and the risk of incident MCI and to evaluate the influence of the apolipoprotein E (APOE) ε4 genotype.

Design, Setting, and Participants  This investigation was a prospective, population-based cohort study of participants in the Mayo Clinic Study of Aging in Olmsted County, Minnesota. Participants 70 years or older who were cognitively normal at baseline were followed up to the outcome of incident MCI. The study dates were April 2006 to June 2016.

Main Outcomes and Measures  At baseline, participants provided information about mentally stimulating activities within 1 year before enrollment into the study. Neurocognitive assessment was conducted at baseline, with evaluations at 15-month intervals. Cognitive diagnosis was made by an expert consensus panel based on published criteria. Hazard ratios (HRs) and 95% CIs were calculated using Cox proportional hazards regression models after adjusting for sex, age, and educational level.

Results  The final cohort consisted of 1929 cognitively normal persons (median age at baseline, 77 years [interquartile range, 74-82 years]; 50.4% [n = 973] female) who were followed up to the outcome of incident MCI. During a median follow-up period of 4.0 years, it was observed that playing games (HR, 0.78; 95% CI, 0.65-0.95) and engaging in craft activities (HR, 0.72; 95% CI, 0.57-0.90), computer use (HR, 0.70; 95% CI, 0.57-0.85), and social activities (HR, 0.77; 95% CI, 0.63-0.94) were associated with a decreased risk of incident MCI. In a stratified analysis by APOE ε4 carrier status, the data point toward the lowest risk of incident MCI for APOE ɛ4 noncarriers who engage in mentally stimulating activities (eg, computer use: HR, 0.73; 95% CI, 0.58-0.92) and toward the highest risk of incident MCI for APOE ɛ4 carriers who do not engage in mentally stimulating activities (eg, no computer use: HR, 1.74; 95% CI, 1.33-2.27).

Conclusions and Relevance  Cognitively normal elderly individuals who engage in specific mentally stimulating activities even in late life have a decreased risk of incident MCI. The associations may vary by APOE ε4 carrier status.

Posted on

Soft robotic sleeve developed to aid failing hearts

A soft robotic sleeve placed around the heart in a pig model of acute heart failure. The actuators embedded in the sleeve support heart function by mimicking the outer heart muscles that induce the heart to beat. (credit: Harvard SEAS)

An international team of scientists has developed a soft robotic sleeve that can be implanted on the external surface of the heart to restore blood circulation in pigs (and possibly humans in the future) whose hearts have stopped beating.

The device is a silicone-based system with two layers of actuators: one that squeezes circumferentially and one that squeezes diagonally, both designed to mimic the movement of healthy hearts when they beat.

Heart failure affects 41 million people worldwide. The concept of an artificial pump that aids cardiac function is not new and has been employed clinically with ventricular assist devices (VADs). “I’ve been implanting VADs in patients for a long time,” says Frank Pigula, MD, a former professor at Harvard and Boston Children’s Hospital who is co-corresponding author of an open-access paper published Jan. 18 in Science Translational Medicine. Pigula is now chief of pediatric cardiac surgery at Norton Children’s Hospital in Louisville, Kentucky.

The current-generation VADs systems directly expose a patient’s blood to artificial materials such as tubing and rotors. When blood touches a device’s components, it has a tendency to clot, which could lead to heart attacks, pulmonary embolism, strokes, or other complications, Pigula explained. To prevent clots from forming, it’s necessary to use blood thinning and anticoagulatory medications such as heparin and warfarin, which are complication-prone.

How to make a soft robotic sleeve for a human heart

That’s where the soft robotic sleeve developed by Pigula’s team comes in.

In vitro demonstration of the pumping of the soft robotic sleeve (credit: Ellen Roche/Harvard University)

The robotic sleeve can be implanted on the external surface of the heart, preventing circulating blood from ever coming in contact with the device’s components, possibly eliminating or decreasing the need for anticoagulatory drugs.

In designing the soft robotic sleeve, the researchers were inspired by the structure and movement of the heart itself. The two actuators are formed in thin layers of silicone that are layered on each other in roughly the same orientation as the muscle fibers in the heart. The sleeve is about half a millimeter thick, which is about the width of 5–10 human hairs.

To adhere the sleeve to pig hearts, the researchers used an FDA-approved adherent on the apex, or tip, of the heart. However, the researchers found this caused severe inflammation at the point of adherence, which could interfere with the ability to implant the sleeve for long periods of time. So the researchers employed a gel to adhere the sleeve to the heart, a technique that lessened the inflammation.

In vivo demonstration of cardiac assist in a porcine model of acute heart failure (credit: Ellen Rouche/Harvard SEAS)

To test the sleeves, the scientists implanted them on the hearts of pigs and induced acute heart failure, resulting in a 50–60% drop in cardiac output. Turning on the sleeve restored 97% of the original cardiac output.

“The soft robotic actuators are essentially artificial muscles,” says Nikolay Vasilyev, MD, a staff scientist in cardiac surgery research at Boston Children’s Hospital and co-author on the recent study. “In this sense, the robotic sleeve mimics both ventricles of the heart.”

The soft robotic heart sleeve also contains sophisticated sensing abilities that measure pressure at specific points on the heart’s surface.

“This work represents an exciting proof-of-concept result for this soft robot, demonstrating that it can safely interact with soft tissue and lead to improvements in cardiac function,” said Conor Walsh, senior author of the paper and the John L. Loeb Associate Professor of Engineering and Applied Sciences at SEAS and Core Faculty Member at Harvard’s Wyss Institute. “We envision many other future applications where such devices can deliver mechanotherapy both inside and outside of the body.”

However, “the human body is remarkably good at detecting foreign materials and mounting immune responses to them, so it will be tricky business to find a biologically inert material that will not, over the long run, scar the tissues it’s physically associated with,” according to a physician who was not involved in the research. “This is less critical for non-vital organs like soft-tissue silicone implants, but a thin layer of scar tissue around the heart could have serious implications for the stiffness, structural integrity, and function of native heart tissue.”

The research was a collaboration between Harvard’s SEAS and Wyss Institute, Boston Children’s Hospital, National University of Ireland,  Technische Universität München, Boston Children’s Hospital, University of Leeds, University of Central Florida, Royal College of Surgeons in Ireland, Trinity College Dublin, UCLA, and University of Louisville.

The work was supported by the Translational Research Program grant from Boston Children’s Hospital, a Director’s Challenge Cross-Platform grant from the Wyss Institute for Biologically Inspired Engineering, Harvard School of Engineering and Applied Sciences, and Science Foundation Ireland.

Abstract of Soft robotic sleeve supports heart function

There is much interest in form-fitting, low-modulus, implantable devices or soft robots that can mimic or assist in complex biological functions such as the contraction of heart muscle. We present a soft robotic sleeve that is implanted around the heart and actively compresses and twists to act as a cardiac ventricular assist device. The sleeve does not contact blood, obviating the need for anticoagulation therapy or blood thinners, and reduces complications with current ventricular assist devices, such as clotting and infection. Our approach used a biologically inspired design to orient individual contracting elements or actuators in a layered helical and circumferential fashion, mimicking the orientation of the outer two muscle layers of the mammalian heart. The resulting implantable soft robot mimicked the form and function of the native heart, with a stiffness value of the same order of magnitude as that of the heart tissue. We demonstrated feasibility of this soft sleeve device for supporting heart function in a porcine model of acute heart failure. The soft robotic sleeve can be customized to patient-specific needs and may have the potential to act as a bridge to transplant for patients with heart failure.

Posted on

Scientists use stem cells to create human/pig chimera embryos

This photograph shows injection of human induced pluripotent stem* (iPS) cells into a pig blastocyst (pre-embryo). A laser beam (green circle with a red cross inside) was used to perforate an opening in the outer membrane (Zona Pellucida) of the pig blastocyst to allow easy access of an injection needle delivering human iPS cells. (credit: Courtesy of Juan Carlos Izpisua Belmonte)

In an open-access paper published online January 26, 2017 in the journal Cell, Salk Institute researchers report breakthroughs on multiple fronts in the race to integrate stem cells from one species into the early-stage development of another species (or chimeras**).

Scientists are still struggling to coax stem cells growing in Petri dishes to become fully functional specialized adult cells, the researchers report. “The ultimate goal is to grow functional and transplantable tissue or organs, but we are far away from that,” says lead investigator Juan Carlos Izpisua Belmonte, a professor in the Salk Institute of Biological Studies’ Gene Expression Laboratory. But their efforts to grow the first embryos containing cells from humans and pigs may someday provide a means of growing human cells, tissues, and organs for regenerative medicine. “This is an important first step.”

Meanwhile, the experiments are helping scientists understand how human stem cells grow and specialize, which may offer insights into disease onset or help establish new drug-testing platforms.

How to grow a chimera

In previous research, Izpisua Belmonte and Salk Institute staff scientist Jun Wu created a rat/mouse chimera by introducing rat cells into mouse embryos and letting the cells mature. The purpose was to find out whether rat-derived cells could rescue severe developmental defects in mouse pancreas, heart or lung. To do that, they used CRISPR genome editing tools to delete critical developmental genes in mouse egg cells. Remarkably, rat cells populated those mouse organs, filling in for the defective mouse cells.

Here, cells derived from rat pluripotent stem cells were enriched in the developing heart of a genetically modified mouse embryo. (credit: Salk Institute)

For instance, in a given cell, the researchers would delete a single gene critical for the development of an organ, such as the heart, pancreas, or eye. Then they introduced rat stem cells into the embryos to see if they would fill the open niche. “The rat cells have a functional copy of the missing mouse gene, so they can outcompete mouse cells in occupying the emptied developmental organ niches,” says Wu. As the organism matured, the rat cells filled in where mouse cells could not, forming the functional tissues of the organism’s heart, eye, or pancreas.

Amazingly, in one experiment, rat cells also grew to form a gall bladder in the mouse, even though rats themselves stopped developing this organ over the 18 million years since rats and mice separated evolutionarily. “This suggests that the reason a rat does not generate a gall bladder is not because it cannot, but because the potential has been hidden by a rat-specific developmental program,” says Wu. “The microenvironment has evolved through millions of years to choose a program that defines a rat.”

Introducing human cells in pig embryos

The team’s next step was to introduce human cells into an organism. Experiments with cow embryos were difficult and costly. They decided to use pig embryos as hosts because the size of these animals’ organs more closely resembles those of humans than those of mice. The effort required completing studies with 1,500 pig embryos and involved the contributions of more than 40 people, including pig farmers, over a four-year period. “We underestimated the effort involved,” says Izpisua Belmonte. “This required a tour de force.”

Salk scientists advance stem-cell and genome-editing technologies to help researchers study evolution and disease, test therapeutic drugs and possibly grow transplantable organs. Here, human iPS cells (green) contributed to a developing heart of 4-week-old pig embryo. (credit: Salk Institute)

The human cells survived and formed a human/pig chimera embryo. Embryos were implanted in sows and allowed to develop for between three and four weeks. “This is long enough for us to try to understand how the human and pig cells mix together early on without raising ethical concerns about mature chimeric animals,” says Izpisua Belmonte.

To ensure that, the human cells did not become precursors of brain cells that can grow into the central nervous system. Rather, they were developing into muscle cells and precursors of other organs. “At this point, we wanted to know whether human cells can contribute at all to address the ‘yes or no’ question,” he says. “Now that we know the answer is yes, our next challenge is to improve efficiency and guide the human cells into forming a particular organ in pigs.”

To do this, the researchers are using CRISPR to perform genome editing on the pig genome, as they did with mice, to open gaps that human cells can fill in. That work is in progress.

“Of course, the ultimate goal of chimeric research is to learn whether we can use stem-cell and gene-editing technologies to generate genetically matched human tissues and organs, and we are very optimistic that continued work will lead to eventual success,” says Izpisua Belmonte. “But in the process we are gaining a better understanding of species evolution as well as human embryogenesis and disease that is difficult to get in other ways.”

Other authors included scientists at the University of Murcia Campus de Espinardo, the University of California, Davis; the Universidad Católica San Antonio de Murcia; Clinica Centro Fundación Pedro Guillén; and the Hospital Clinic of Barcelona. The work was funded by The Fundación Séneca; a University of California, Davis, Academic Senate New Research Grant; the Universidad Católica San Antonio de Murcia; Fundacion Dr. Pedro Guillen; the G. Harold and Leila Y. Mathers Charitable Foundation; and The Moxie Foundation.

* Induced pluripotent stem cells (also known as iPS cells or iPSCs) are a type of pluripotent stem cell that can be generated directly from adult cells. They have the capacity to mature into many different types of cells and therefore can contribute to the formation of multiple organs or distinct cell lineages, but not all possible lineages. This developmental capacity differs from naïve stem cells, which can develop into any given cell type.

** An interspecies chimera is an organism containing cells from different species. The word “chimera” originally described mythological creatures or deities in polytheistic religions. In science, interspecies chimeras have emerged as valuable basic research tools with potential for future clinical applications.


Salk Institute | New findings highlight promise of chimeric organisms for science and medicine

Abstract of Interspecies Chimerism with Mammalian Pluripotent Stem Cells

Interspecies blastocyst complementation enables organ-specific enrichment of xenogenic pluripotent stem cell (PSC) derivatives. Here, we establish a versatile blastocyst complementation platform based on CRISPR-Cas9-mediated zygote genome editing and show enrichment of rat PSC-derivatives in several tissues of gene-edited organogenesis-disabled mice. Besides gaining insights into species evolution, embryogenesis, and human disease, interspecies blastocyst complementation might allow human organ generation in animals whose organ size, anatomy, and physiology are closer to humans. To date, however, whether human PSCs (hPSCs) can contribute to chimera formation in non-rodent species remains unknown. We systematically evaluate the chimeric competency of several types of hPSCs using a more diversified clade of mammals, the ungulates. We find that naïve hPSCs robustly engraft in both pig and cattle pre-implantation blastocysts but show limited contribution to post-implantation pig embryos. Instead, an intermediate hPSC type exhibits higher degree of chimerism and is able to generate differentiated progenies in post-implantation pig embryos.