Thursday, July 21st, 2016

Time	Event
12:00a	Reducing wait times at the doctor’s office Ever waited entirely too long at your doctor’s office for an appointment to start? The long wait may soon be over: An MIT spinout’s schedule-optimizing software that gets more patients seen more quickly could soon be used by tens of thousands of health care providers across the country, after a recent acquisition by a major health care services company. Arsenal Health has developed a schedule-optimization service for health care providers that began as a pitch at MIT Hacking Medicine, a hackathon that aims to solve problems in health care. The service analyzes scheduling and other data to predict which patients might not show up to appointments. Health-care providers can then double-book over those potential no-show patients or prioritize their outreach as needed. Double-booking is often a necessary evil for health-care providers who see high volumes of patients. Inevitably, some of those patients cancel too late or don’t show to appointments, which is costly and a time-waster. Front-end staff can get ahead of this by double-booking, allowing them to move patients around to make up for the no-shows and enabling more patients to be seen overall. But if both parties show up for a single appointment slot, one person waits longer. Arsenal Health’s service, on the other hand, is a type of “targeted double-booking” that can predict that one patient won’t show, much more accurately than when administrative staff double-book manually, says co-founder and former CEO Chris Moses ’10, now director of product innovation at athenahealth. “This means improved patient access and availability, and improved provider productivity by making their scheduling more open,” he says. Arsenal Health was acquired in April by athenahealth, which provides cloud-based, network-enabled services and apps for more than 78,000 health-care providers nationwide. Currently, more than 800 of those providers use Arsenal Health’s technology. Win-win-win Arsenal Health’s solution gathers and analyzes clinical, administrative, and scheduling data to find trends of when and why patients cancel. (Such factors as potential sickness or bad weather, for instance, may be out of a patient’s control.) Using that information, the software uses predictive modeling that determines if a certain patient will or won’t show on a particular day and time. When front-desk staff are searching for open appointments in athenahealth, the scheduling tool flags the patients that might not show, so they can double-book those slots, called “smart open slots.” It also lists potential no-shows in a call list in a web app so that front-desk staff can reach out to those targeted patients, which has been shown to decrease no-show rates, Moses says. According to Moses, manually double-booking patients without Arsenal Health is about 20 to 30 percent accurate, meaning both patients show up to an appointment about 70 to 80 percent of the time. The Arsenal Health solution, however, is about 75 percent accurate, with both patients showing up only about a quarter of the time, which cuts waiting time and improves patient satisfaction, Moses says. In some cases, Arsenal Health’s service has also increased the number of new patients seen by providers, Moses says. On average, he says, a provider using Arsenal Health’s schedule optimization gains a couple more patients each month, and some have gained an additional 30 or 40 patients each month. “That’s important for medical groups and primary-care practices that work in these hospital systems, because new patients equals more revenue for the hospitals, while you’re improving patient experience,” Moses says. By decreasing no-show rates and increasing new patient numbers, the software has boosted revenue of participating providers by roughly $700 per month, according to Arsenal Health. “So it’s a win-win-win on all sides,” Moses says. Hack to acquisition Two years after graduating MIT, Moses attended the 2012 MIT Hacking Medicine event, which was co-organized by a friend. During the weekend-long hackathon, MIT postdoc Gabriel Belfort pitched the no-show problem that his wife, a pediatrician, often complained about. “I thought that was one of biggest problems, one of the realest needs I heard that weekend,” Moses says. Moses and Belfort joined up with another physician, Donald Misquitta, who was also a data scientist, and MIT PhD engineering student Andrea Ippolito, a Hacking Medicine co-founder. They formed a team dedicated to developing commercial software to solve the no-show problem, with Moses becoming the founding full-time employee. That summer, the four-person team was accepted into the Founders’ Skills Accelerator at the Martin Trust Center for MIT Entrepreneurship, “which was an amazing opportunity,” Moses says. Among other things, the program provided a “board” of directors, made up of seasoned entrepreneurs, whom the team submitted milestones to on a monthly basis. “The board would say, ‘We’re not giving you $20,000 upfront. You have to earn it,’” Moses says. “That was a really cool structure that helped create accountability really early on.” In the summer of 2012, the team entered the Healthbox startup accelerator in Boston, where they partnered with Steward Health Care, a major hospital system in Massachusetts. As luck would have it, Steward was an athenahealth enterprise customer, and agreed to work with Arsenal Health and make their data available for training the initial predictive models. “It was a match made in heaven,” Moses says. The team looked through five years’ worth of scheduling data from 17 offices across several Steward hospitals, and found that of 700,000 appointments, there were 30,000 no-shows — which confirmed there was a real problem, Moses says. Using athenahealth’s data, they built their first prototype that predicted future no-shows at Steward pilot hospitals. After that, Steward became Arsenal’s first paying customer. “When their paycheck hit the bank, we became immediately profitable given our small team size,” Moses says. In 2014, athenahealth launched its “More Disruption Please” accelerator program, recruiting Arsenal Health as its first investment — which ultimately led to a strong partnership and the acquisition in April. Now Moses is working with a team to implement the scheduling service across athenahealth’s entire network. “The growth opportunity is amazing,” Moses says. “That’s something we could never do as an independent company.” The acquisition also means more development on predictive modeling and more research into what really causes patients no-shows. “[We’ll] look at the problems across athenahealth’s customer base — whether they’re internal efficiency problems or external customer problems,” Moses says. “Using data to better improve customers’ care.” (Comment on this)
2:00p	Scientists program cells to remember and respond to series of stimuli Synthetic biology allows researchers to program cells to perform novel functions such as fluorescing in response to a particular chemical or producing drugs in response to disease markers. In a step toward devising much more complex cellular circuits, MIT engineers have now programmed cells to remember and respond to a series of events.   These cells can remember, in the correct order, up to three different inputs, but this approach should be scalable to incorporate many more stimuli, the researchers say. Using this system, scientists can track cellular events that occur in a particular order, create environmental sensors that store complex histories, or program cellular trajectories. “You can build very complex computing systems if you integrate the element of memory together with computation,” says Timothy Lu, an associate professor of electrical engineering and computer science and of biological engineering, and head of the Synthetic Biology Group at MIT’s Research Laboratory of Electronics. This approach allows scientists to create biological “state machines” — devices that exist in different states depending on the identities and orders of inputs they receive. The researchers also created software that helps users design circuits that implement state machines with different behaviors, which can then be tested in cells. Lu is the senior author of the new study, which appears in the 22 July issue of Science. Nathaniel Roquet, an MIT and Harvard graduate student, is the paper’s lead author. Other authors on the paper include Scott Aaronson, an associate professor of electrical engineering and computer science, recent MIT graduate Ava Soleimany, and recent Wellesley College graduate Alyssa Ferris. Long-term memory In 2013, Lu and colleagues designed cell circuits that could perform a logic function and then store a memory of the event by encoding it in their DNA. The state machine circuits that they designed in the new paper rely on enzymes called recombinases. When activated by a specific input in the cell, such as a chemical signal, recombinases either delete or invert a particular stretch of DNA, depending on the orientation of two DNA target sequences known as recognition sites. The stretch of DNA between those sites may contain recognition sites for other recombinases that respond to different inputs. Flipping or deleting those sites alters what will happen to the DNA if a second or third recombinase is later activated. Therefore, a cell’s history can be determined by sequencing its DNA. In the simplest version of this system, with just two inputs, there are five possible states for the circuit: states corresponding to neither input, input A only, input B only, A followed by B, and B followed by A. The researchers also designed and built circuits that record three inputs, in which 16 states are possible. For this study, the researchers programmed E. coli cells to respond to substances commonly used in lab experiments, including ATc (an analogue of the antibiotic tetracycline), a sugar called arabinose, and a chemical called DAPG. However, for medical or environmental applications, the recombinases could be re-engineered to respond to other conditions such as acidity or the presence of specific transcription factors (proteins that control gene expression). Gene control After creating circuits that could record events, the researchers then incorporated genes into the array of recombinase binding sites, along with genetic regulatory elements. In these circuits, when recombinases rearrange the DNA, the circuits not only record information but also control which genes get turned on or off. The researchers tested this approach with three genes that code for different fluorescent proteins — green, red, and blue, constructing a circuit that expressed a different combination of the fluorescent proteins for each identity and order of two inputs. For example, when cells carrying this circuit recieved input A followed by input B they fluoresced red and green, while cells that recieved B before A fluoresced red and blue. Lu’s lab now hopes to use this approach to study cellular processes that are controlled by a series of events, such as the appearance of cytokines or other signaling molecules, or the activation of certain genes. “This idea that we can record and respond to not just combinations of biological events but also their orders opens up a lot of potential applications. A lot is known about what factors regulate differentiation of specific cell types or lead to the progression of certain diseases, but not much is known about the temporal organization of those factors. That’s one of the areas we hope to dive into with our device,” Roquet says. For example, scientists could use this technique to follow the trajectory of stem cells or other immature cells into differentiated, mature cell types. They could also follow the progression of diseases such as cancer. A recent study has shown that the order in which cancer-causing mutations are acquired can determine the behavior of the disease, including how cancer cells respond to drugs and develop into tumors. Furthermore, engineers could use the state machine platform developed here to program cell functions and differentiation pathways.  The MIT study represents “a new benchmark in the use of living cells to perform computation and to record information,” says Tom Ellis, a senior lecturer at the Centre for Synthetic Biology at Imperial College London. “These recombinase-based state machines open up the possibility of cells being engineered to become recorders of temporal information about their environment, and they can be built to lead the cells to take actions in response to the appropriate string of inputs,” says Ellis, who was not involved in the research. “It's an excellent paper that puts these recombinase-based switches to good use.” (Comment on this)
2:00p	Borrowing from pastry chefs, engineers create nanolayered composites Adapting an old trick used for centuries by both metalsmiths and pastry makers, a team of researchers at MIT has found a way to efficiently create composite materials containing hundreds of layers that are just atoms thick but span the full width of the material. The discovery could open up wide-ranging possibilities for designing new, easy-to-manufacture composites for optical devices, electronic systems, and high-tech materials. The work is described this week in a paper in Science by Michael Strano, the Carbon P. Dubbs Professor in Chemical Engineering; postdoc Pingwei Liu; and 11 other MIT students, postdocs, and professors. Materials such as graphene, a two-dimensional form of pure carbon, and carbon nanotubes, tiny cylinders that are essentially rolled-up graphene, are “some of the strongest, hardest materials we have available,” says Strano, because their atoms are held together entirely by carbon-carbon bonds, which are “the strongest nature gives us” for chemical bonds to work with. So, researchers have been searching for ways of using these nanomaterials to add great strength to composite materials, much the way steel bars are used to reinforce concrete. The biggest obstacle has been finding ways to embed these materials within a matrix of another material in an orderly way. These tiny sheets and tubes have a strong tendency to clump together, so just stirring them into a batch of liquid resin before it sets doesn’t work at all. The MIT team’s insight was in finding a way to create large numbers of layers, stacked in a perfectly orderly way, without having to stack each layer individually. Although the process is more complex than it sounds, at the heart of it is a technique similar to that used to make ultrastrong steel sword blades, as well as the puff pastry that’s in baklava and napoleons. A layer of material — be it steel, dough, or graphene — is spread out flat. Then, the material is doubled over on itself, pounded or rolled out, and then doubled over again, and again, and again. With each fold, the number of layers doubles, thus producing an exponential increase in the layering. Just 20 simple folds would produce more than a million perfectly aligned layers. Now, it doesn’t work out exactly that way on the nanoscale. In this research, rather than folding the material, the team cut the whole block — itself consisting of alternating layers of graphene and the composite material — into quarters, and then slid one quarter on top of another, quadrupling the number of layers, and then repeating the process. But the result was the same: a uniform stack of layers, quickly produced, and already embedded in the matrix material, in this case polycarbonate, to form a composite. In their proof-of-concept tests, the MIT team produced composites with up to 320 layers of graphene embedded in them. They were able to demonstrate that even though the total amount of the graphene added to the material was minuscule — less than 1/10 of a percent by weight — it led to a clear-cut improvement in overall strength. “The graphene has an effectively infinite aspect ratio,” Strano says, since it is infinitesimally thin yet can span sizes large enough to be seen and handled. “It can span two dimensions of the material,” even though it is only nanometers thick. Graphene and a handful of other known 2-D materials are “the only known materials that can do that,” he says. The team also found a way to make structured fibers from graphene, potentially enabling the creation of yarns and fabrics with embedded electronic functions, as well as yet another class of composites. The method uses a shearing mechanism, somewhat like a cheese slicer, to peel off layers of graphene in a way that causes them to roll up into a scroll-like shape, technically known as an Archimedean spiral. That could overcome one of the biggest drawbacks of graphene and nanotubes, in terms of their ability to be woven into long fibers: their extreme slipperiness. Because they are so perfectly smooth, strands slip past each other instead of sticking together in a bundle. And the new scrolled strands not only overcome that problem, they are also extremely stretchy, unlike other super-strong materials such as Kevlar. That means they might lend themselves to being woven into protective materials that could “give” without breaking. One unexpected feature of the new layered composites, Strano says, is that the graphene layers, which are extremely electrically conductive, maintain their continuity all the way across their composite sample without any short-circuiting to the adjacent layers. So, for example, simply inserting an electrical probe into the stack to a certain precise depth would make it possible to uniquely “address” any one of the hundreds of layers. This could ultimately lead to new kinds of complex multilayered electronics, he says. This paper “describes a rather unique and creative way to make composites using large area graphene films,” says Angelos Kyrlidis, research and development manager for graphenes at Cabot Corporation who was not involved with this work. He adds, “This work assembles the composites from chemical vapor deposition graphene, where a very high aspect ratio can be obtained, while still maintaining many of the features and properties of the single layer graphene. … It would be quite interesting to evaluate in a broader range of polymers, such as thermosets and also other thermoplastics.” The research was supported by the U.S. Army Research Office through the Institute for Soldier Nanotechnologies at MIT. (Comment on this)

<< Previous Day

2016/07/21 [Calendar]

Next Day >>