Koch Foundation

NOVA Science Trust Logo

Hervé This, co-creator of molecular gastronomy, holds up one of his creations for inspection. This is now on to a new, more challenging way of cooking.

Synthetic Food: Better Cooking Through Chemistry

<div class="related"></div>
<p>
</p>
<p>
 Just as in art, music, dance, and literature, trends in cooking go in and out of vogue. The latest movement to sweep the world’s kitchens is molecular gastronomy, essentially the application of the scientific method to food preparation. The basic idea behind molecular gastronomy is to determine what happens when food cooks, how its physical and chemical properties change, how different methods modify its taste, smell, and texture, and then use this new understanding to create novel dishes. Dozens of high-end restaurants worldwide now serve innovative delicacies such as snail porridge, bacon-and-egg ice cream, and a wide range of edible foams, gels, and “spherificated” foods.
</p>
<p>
 But when I ask Hervé This (pronounced “Tees”), one of co-developers of molecular gastronomy, about cooking’s current cutting-edge via Skype, he responds: “
 <i>
  En bref, gastronomie molecular est passé,
 </i>
 ” with a shake of his unruly shock of ashen hair. Instead, This wants to go one step further, not just to prepare meals inspired by chemical and physical insights, but “to make food directly from the basic constituent chemicals themselves—the individual flavor notes that comprise dishes,” he says. “This next trend in cooking is what we call note-by-note cuisine.”
</p>
<p>
 This, a French physical chemist, Nicholas Kurti, the late Hungarian-born physicist, and others began detailed research into molecular gastronomy in the mid-1990s. The application in the kitchen of new techniques—many developed by the food-processing industry—proved useful and provocative, particularly since it led to the rise of an international culinary movement that has identified and exploited high-tech cooking methods.
</p>
<p>
 In the late 1990s and early 2000s, the term “molecular gastronomy” grew to refer to a distinctive set of fairly exotic processes that often rely on costly machines used in “modernist” restaurants and kitchens to create innovative dishes. Chefs in those kitchens don’t just chop and stir, they use tools such as siphons, water baths, liquid nitrogen, and rotary evaporators. Their ingredient lists sound like something out of a biochemistry textbook: sodium alginate, agar-agar and other plant gums, and enzymes such as transglutaminase.
</p>
<p>
 Molecular gastronomy is now widespread enough to be considered established, making it something to be studied and taught in academia, as This does at AgroParisTech. It’s no longer a movement operating on the sidelines. “That time was all wonderful,” This declares, gesturing at me in our video chat from his home office in Paris, “but that was the past.”
</p>
<p>
 It seems the alchemist has lost interest in the Philosopher’s Stone.
</p>
<h2>
 Vive Note à Note
</h2>
<p>
 But This has a new muse. He leans forward, moving on to what currently drives him: “note-by-note” cuisine, the ultimate in cooking from scratch. Rather than pulling ingredients from the cupboard or refrigerator, This now wants chefs to grab them from the chemical cabinet. If successful, note-by-note could reshape the culinary landscape, and, if This’s long-term vision pans out, do away with food distribution as we know it.
</p>
<figure class="richtext-image full-width">
<img alt="" class="richtext-image full-width" src="https://staging.pbs.org/wgbh/nova/media/images/note-a-note-champagne.width-990_mjY7UfV.jpg"/>
<figcaption>A note-by-note champagne
        
    </figcaption>
</figure>
<p>
 This’s concept, which he discusses in his new book,
 <a href="http://www.amazon.fr/dp/2701164192">
<i>
   La cuisine note à note
  </i>
</a>
 , (an English translation of which is to be published by Columbia University Press), is to identify and study the fundamental chemical constituents of dishes—the myriad molecules in the meals we eat—and then reconstruct the essence of those recipes using a selection of the compounds that create a dish’s sensory experience. Beyond recreating something like (but unlike) the foods we know, he envisions the eventual possibility of engineering never-eaten-before creations.
</p>
<p>
 This says that a note-by-note chef would design dishes by composing odors and tastes, colors, shapes, and consistencies—even nutritional values—entirely from the mixing of “raw” compounds. Ingredients can include water, ethanol, sucrose, amino acids, lipids, and so forth. “Using chemical compounds opens up billions of new possibilities,” he says. “It’s like a painter using a palette of primary colors or a musician composing note by note.”
</p>
<h2>
 Promoting New Notes
</h2>
<p>
 So far, many note-by-note recipes seem to be relatively simple, liquid—or at least mostly smooth-textured—and often alcoholic. A faux orange cocktail, for example, might include ethanol, citric acid, fructose, glucose, and beta-carotene, and a note-by-note wine might be mixed up from water, anthocyanins (for color), sugars, ethanol, and amino acids (for flavor) as well as glycerol, phenols, quinones, and organic acids.
</p>
<figure class="richtext-image full-width">
<img alt="" class="richtext-image full-width" src="https://staging.pbs.org/wgbh/nova/media/images/note-a-n-oeuf-cordon-bleu.width-990_wh3jHaK.jpg"/>
<figcaption>This egg dish was one of many note-by-note recipes recently cooked and served the cooking school Le Cordon Bleu Paris.
        
    </figcaption>
</figure>
<p>
 Note-by-note recipes tend to look more at home in a lab notebook than a kitchen binder. Consider this early attempt to make a fully synthetic note-by-note sauce, basically a wine-less wine sauce for a butter-poached lobster fricassee, annotated here:
</p>
<blockquote>
<p>
 Melt 100 g of glucose and 20 g of tartaric acid (one of the main acids in wine) in 200 ml of water. Add 2 g of polyphenol, which has been extracted from grape juice using reverse osmosis filtering. Boil and add sodium chloride (table salt) and piperine—the pungent agent of black pepper. Bind the sauce with amylose, a polysaccaride, one of the two components of starch. Remove the preparation from off the heat and stir in 50 g of triacylglycerol (a triglyceride or fat and oil component). Serve as a sauce.
</p>
</blockquote>
<p>
 This first proposed note-by-note cuisine after he was experimenting with adding flavors to food and drink to improve the taste. “As I mentioned in the introductory 1994
 <i>
  Scientific American
 </i>
 paper on note-by-note,” This recalls, “you might want to add the aromatic molecule, vanillin, to a cheaper whiskey before serving it to guests because vanilla is found in some more expensive whiskeys.” Before long, he says, “I was thinking about making dishes directly from the compounds alone.”
</p>
<div class="pullquote">
<p>
 “I’m still waiting for a chef to open a restaurant that clearly serves note-by-note dishes.”
</p>
</div>
<p>
 This vividly remembers the excitement of his early lectures on the note-by-note concept, but admits that promoting such a radical philosophy turned out to be tougher than he thought. He mentions that he gets no direct remuneration from championing the cause, and that it comes with its own costs. “I once got a couple of death threats from opponents of agro-businesses…”
</p>
<p>
 Following his initial efforts, This cut back on proselytizing for a while. But a few years ago, he began to step up his advocacy. “By then I felt more comfortable with the procedures,” he says. “I teamed up with my friend, chef Pierre Gagnaire, and we worked for about six months to a year to develop dishes before presenting the first ‘note à note’ meal in Hong Kong in 2008.” (Gagnaire is a three-star Michelin chef with an eponymous restaurant in Paris.) The first-ever note-by-note dish featured jelly pearls that tasted like apple, iced granite—a slightly crunchy sorbet—with a lemon-like taste, and a wafer-thin “glucose caramel” strip. No apple, lemon, or caramel were used to flavor the dish.
</p>
<figure class="richtext-image full-width">
<img alt="" class="richtext-image full-width" src="https://staging.pbs.org/wgbh/nova/media/images/note-a-note-chick-corea.width-990_TSWVgib.jpg"/>
<figcaption>A chef plates "Chick Corea," a note-by-note creation by Pierre Gagnaire, who named the dish after the jazz-fusion pianist.
        
    </figcaption>
</figure>
<p>
 Every year, This, together with chefs and students at Le Cordon Bleu Paris, prepare a gala note-by-note dinner, and he hosts other tasting events in Hong Kong and Montreal to promulgate the practice. He and two AgroParisTech colleagues recently held the inaugural note-by-note cooking contest.
</p>
<h2>
 Beyond the Kitchen
</h2>
<p>
 Like many culinary movements, such as organic cooking and local sourcing, This’s motivations for developing note-by-note cuisine aren’t entirely gastronomic. “We face an energy crisis, and as energy costs rise, transportation costs rise. But by extracting the water near the food source, you can avoid the costs of transporting the water in food,” he says. “Say you pulp some carrots, remove the cellulose, and then separate the remaining liquid using reverse osmosis into water and a ‘not-water’ fraction. People are just thinking about uses for cellulose, and the non-water, non-cellulose remnants would be something new in the kitchen that would be easy to use. You could divide it further if you desire.”
</p>
<p>
 Fractionations and water extraction would also avoid spoilage losses, This adds, which accounts for 20-40% of global food supplies. The approach could “feed a lot more people with the same food supply” and would help farmers remain profitable, he says.
</p>
<p>
 “This’s articles give a quite good analysis of his arguments,” notes Jens Risbo, a food scientist at the University of Copenhagen, “though I don’t share his conclusions. I would, for example, expect that the energy used to purify the extracts would exceed what you could gain by avoiding shipping water, but I understand his intentions.”
</p>
<p>
 “There’s no doubt that from an artistic point of view, note-by-note definitely moves boundaries, freeing up some long-held limitations in making food creations, and enabling practitioners to explore a new part of the cooking space,” Risbo says. “I guess I just don’t think it’s going to be something that people will do much of. It will always be limited. Developing the right chemical compounds to recreate something like food is not that easy to do,” he continues. “It’s a significant challenge that will require new understanding. And nature does a marvelous job of that already.”
</p>
<div class="ad"></div>
<h2>
 Mixed Results
</h2>
<p>
 In the culinary world, This’s concept gets a mixed reception. “This’s vision is provocative and potentially instructive, which is no surprise as he’s something of a visionary in opening up food science to fine cooking,” says Nathan Myhrvold, technologist and principal author of the six-volume, 2,400-page
 <i>
  Modernist Cuisine
 </i>
 . “I would venture that a series of simple dishes conceived using a sparse pallet of flavors could be highly enlightening,” he says. “You might, for instance, be able to get a better idea of how the dominant and lesser flavors interact as they play on your tongue because you’re dealing with fewer of them.” It’s similar to how scientists test their hypotheses—by limiting the number of variables to study the desired effect.
</p>
<p>
 “What realistic impacts of note-by-note might have in the long-term is hard to gauge,” Myhrvold says. “But perhaps you might end up buying a beef broth in the supermarket that has no meat, but is believable enough because it contains all the fundamental essences that constitute beef broth.”
</p>
<p>
 Not everyone is as optimistic. Several less-than-positive press reviews have discussed dishes, say an orange cocktail or a fish custard, that at best vaguely remind tasters of the real thing. It seems as if the proponents of note-by-note are still working out the top-notes of particular dishes—the dominant, important flavors, odors, and textures that are absolutely essential to recognizing and savoring them. Results may improve when chefs begin adding more subtle, trace flavor compounds to note-by-note dishes to fill out the sensory arrays. For his part, This admits, with some reservations, that his approach is still in its infancy.
</p>
<p>
 Fortunately, others report much more positive results. Peter Barham, a polymer physicist at the University of Bristol and visiting professor of molecular gastronomy at the Copenhagen’s Royal Veterinary University, writes in a recent e-mail that he sometimes prepares “a special ‘organic’ raspberry sorbet when I have foodies for dinner. It tastes just fine.”
</p>
<p>
 After Barham’s guests have eaten and commented on it, he then reveals that the only thing about the recipe that’s “organic,” save the water, is the organic chemist who made the ingredients. “The point is,” he writes, “to illustrate the fact that nobody can simply from the flavor tell whether a food is organic or not, nor even how it was made—it is all in the story we weave when we serve food!”
</p>
<div class="newsletter"></div>
<h2>
 Establishing
 <i>
  la Technique
 </i>
</h2>
<p>
 In an effort to build up the bona fides of the note-by-note approach, This and his co-workers aren’t just using the technique to invent new dishes, they are also dissecting old recipes. The team has reportedly analyzed more than 450 traditional French sauces, from béchamel to velouté, and found that they follow just 23 general formulas. He reportedly tries to sell chefs on the utility of his method by
 <a href="http://www.finedininglovers.com/stories/food-science-herve-this/">
  generating formulas
 </a>
 entirely at random just to show that they will work.
</p>
<p>
 Of course, some recipes are less successful than others. “In November 2010,” This says, “I met with a group of French chefs to give private lessons on the principles of note-by-note. I bought various products that produce color, consistency, mouth feel, etc., and demonstrated some of the basic combinations and procedures. Then I let them play for a few hours. Nobody was happy with most of the results, but what do you expect? This is the first time they had used ingredients like beta carotene or chlorophyll A.”
</p>
<figure class="richtext-image full-width">
<img alt="" class="richtext-image full-width" src="https://staging.pbs.org/wgbh/nova/media/images/meat-spray.width-990_sDaxwjp.jpg"/>
<figcaption>This note-by-note main course is notable for its "meat spray."
        
    </figcaption>
</figure>
<p>
 “It was the same at the beginning of molecular cuisine,” he says. “It took a while before many chefs grew comfortable with using alginate and calcium to make hydrocolloid gels.”
</p>
<p>
 This admits that “it’ll take a minimum of 10 years” for his ideas to catch on. “I’m really still waiting for a chef to open a restaurant that clearly serves note-by-note dishes.” He predicts that the pioneer will probably be “an old, established chef, one who’s still a rebel.” That chef, he guesses, will probably not be in France, Europe, or the U.S., but perhaps Asia or Australia.
</p>
<p>
 For its part, the food industry is intrigued, but hasn’t embraced This’s latest innovation. He says that he has discussed flavoring with managers at Yoplait and other corporations, and adds that many industry personnel attend his courses. “But the food industry is very conservative,” This warns. “They don’t move toward possible business uses unless they know that the consumer is onboard.”
</p>
<p>
 In the meantime, This is emphasizing a more accessible adaptation of note-by-note. It’s a more realistic method that he calls “practical note-by-note,” one that employs more chemically complex extractions and essences rather than basic compounds. That should make it a lot easier to master. The simplified, beginner’s version is like “playing a kid’s synthesizer keyboard,” he says. “It doesn’t offer all the possible notes, but it’s easy to use so at least you can make some music.”
</p>
<p>
 Photo credits: © Alexandra Boulat/VII/Corbis, Hervé This.
</p>

<p>Chemist and chef Hervé This has something new he’d like you to try.</p>

Flying the Electric Skies

<p>
 The sight of an airplane wing perched atop a big-rig truck—especially one that’s rumbling across a dry lake bed at dawn—might not seem significant to the future of aviation. But when the wing’s 18 small propellers flash into the sun’s first rays like a swarm of manic red butterflies, the spectacle actually provides a good indication of what may be in store for next-generation aircraft.
</p>
<div class="related"></div>
<p>
 Those morning speed runs across Edwards Air Force Base in California represent some of NASA’s first real-world tests of distributed electric aircraft propulsion technology, a budding design concept that could start to revolutionize everything from light planes to regional airliners within a decade or so, says Mark D. Moore, principal researcher for NASA’s Convergent Electric Propulsion Technology Sub-Project at Langley Research Center.
</p>
<div class="pullquote">
<p>
 “DEP could mean a fundamental shift in how we design aircraft.”
</p>
</div>
<p>
 Today, when you see a prop plane or jet flying overhead, it is driven through the air by propellers or fan blades spun by powerful engines that burn fossil fuels. But if distributed electric propulsion (DEP) is successfully developed they could be quieter, more efficient, cleaner, and reliable than conventional piston and turbine engines. They would also make a significant dent in the aviation industry’s carbon footprint, which today is about 2% of all emissions.
</p>
<p>
 The unusual “blown wing” that NASA’s Hybrid-Electric Integrated Systems Test-bed (HEIST) is evaluating features an array of small electrically powered propellers that send multiple streams of high-speed air over the upper surface of the wing to produce unprecedented lift capabilities.
</p>
<p>
 Such battery- or hybrid-electric aero-propulsion technology could also enable very different-looking aircraft to take to the skies. “DEP could mean a fundamental shift in how we design aircraft,” Moore says. Imagine, for instance, a light airplane with a dozen small electric propellers up front rather than just one or two large props. Or picture an airliner that, instead of being powered by the typical engines that hang under the wings, is driven by a dozen smaller ducted fans lining the wings’ trailing edges. In time, DEP could make possible novel and potentially more capable airplane designs.
</p>
<p>
 Everyday electric aircraft are still probably more than a decade or two away, but researchers and engineers from all corners of the aerospace industry have been making surprising progress, from airframe manufacturers including Airbus and Boeing to engine makers such as Siemens, Rolls Royce, and GE to small aero-engineering firms like California-based Joby Aviation, ES Aero, and Zee.Aero. University and government labs are also working the technology, says Ruben Del Rosario, manager of the Fixed Wing Project of the Fundamental Aeronautics Program at NASA’s Glenn Research Center.
</p>
<p>
 “Of course, many other experts still strongly doubt that electric propulsion for larger aircraft will be available for decades to come,” he says. But there seems to have been a sea change recently. “Two or three years ago, there was little hope among experts that you could even build electric motors that were powerful and lightweight enough to do the job for anything but rather small airplanes, but recent advancements in materials and the development of new airframe and motor design and engineering concepts are changing that view.”
</p>
<h2>
 Efficient E-Aircraft
</h2>
<p>
 The main advantage of DEP is that the enhanced airflow over the wings can enable an airplane to fly more efficiently. “If you distribute higher velocity air across the entire wing, you can raise the dynamic pressure over the wing and thus increase lift substantially at low flight speeds,” Moore says. The array of small propellers would, for example, let planes take off and land on shorter runways.
</p>
<p>
 With DEP, wings would only need to be a third of the width as on a conventional aircraft, saving weight and fuel costs, he says. On light aircraft, that’s particularly beneficial, since they are typically relatively large to prevent the plane from stalling, or losing lift at low speeds. But large-area wings are inefficient when an aircraft is cruising at speed because they create considerable aerodynamic drag.
</p>
<div class="ad"></div>
<p>
 A more slender, DEP-equipped wing could also be smoother flying and more fuel efficient, Moore says. It would also be safer—losing one engine mid-flight isn’t as much of a problem when you have a dozen or more. And they would even generate less noise because their tips slice through the air more slowly than standard propellers.
</p>
<p>
 NASA’s experimental HEIST test article, a 31-foot-span, carbon-composite wing section, is mounted on a supporting truss that floats on a vibration-absorbing airbag, says Sean Clark, project engineer at Armstrong Flight Research Center. Combined, the 18 props produce about 300 horsepower, and the wing creates around 3,500 pounds of lift. The ground-test rig serves as a “mobile wind tunnel” that generates data at significantly less cost than a conventional large-scale wind tunnel.
</p>
<h2>
 A Leap Forward
</h2>
<p>
 The HEIST ground tests are part of NASA’s $15-million, three-year Leading Edge Asynchronous Propeller Technology (LEAPTech) program, which is evaluating how DEP can be integrated into airframes and what benefits that integration might produce.
</p>
<p>
 To develop and build the rolling test bed, NASA engineers at Langley and Armstrong partnered with “two small, nimble, enthusiastic firms,” Moore says—Empirical Systems Aerospace, the prime contractor, and Joby Aviation, which built the test rig, wing, motors, and propellers.
</p>
<p>
 NASA’s next step is a project called SCEPTOR in which the wing of a four-seat light aircraft—a twin-engine Italian-built Tecnam P2006T—will be replaced with a DEP wing that incorporates perhaps a dozen electrically driven propellers.
</p>
<p>
 Using an existing airframe will allow researchers to compare the performance of the modified plane with the original. “We’re going to characterize the baseline aircraft while in parallel testing the design of the new DEP wing,” he says. “Then we’ll pull off the old wings and put on the new ones.”
</p>
<p>
 “A very significant portion of this research effort, which is being performed by Sean Clark’s group at Armstrong, is to develop the distributed power systems, the many motors, motor controllers, propulsors and master power controller as well as to understand the complex architecture by which they can all talk to each other,” Moore says.
</p>
<p>
 SCEPTOR’s wing will be optimized for high-speed cruise—around 170 mph—yet still supply enough lift to help prevent stalling on takeoff or landing. The DEP wing is to perform other tricks as well. The speed of each electric propeller can be controlled independently, which allows engineers to tweak the airflow over the wing to cope with fast-changing flight conditions such as wind gusts. When cruising, the propellers nearer the fuselage could be folded back to cut drag, leaving those on the wing tips to do the work.
</p>
<p>
 The re-winged Tecnam is scheduled to start test flights in 2017. “We hope to showcase that DEP can achieve a five-times reduction in energy consumption in high-speed cruise over the baseline aircraft,” he says, adding that the researchers are applying for “X-plane” status for the aircraft, a coveted designation.
</p>
<p>
 “Think of the Tecnam X-plane as a subscale demonstrator. We want to incubate the concept at the general aviation [light plane] level and then scale it up,” Moore says. If the test flights are successful, DEP technology could be incorporated into small aircraft within a decade, especially if battery performance and weight improves as much as experts anticipate.
</p>
<h2>
 New Transportation Solutions
</h2>
<p>
 NASA’s DEP research efforts got their start in 2011, when Moore and his colleagues began studying the possibility of “turning the small airplane into a real mid-range transportation solution instead of a mostly recreational novelty,” he recalls. “An automobile works great up to about 100 miles, while a commercial airliner works fine for from 500 to 1,000 miles, but for affordable, high-speed mobility between 100 to 500 miles, there’s no great transportation solution. Our studies say that distributed electric propulsion would be cost-effective for distances of less than 600 miles.”
</p>
<p>
 DEP coupled with self-driving auto-pilot systems that would handle takeoffs and landings could provide dramatically higher-speed, more affordable travel than cars but with car-like ease of use, he says. But only a fully “bulletproof-safe” flight-control system could ensure nearly accident-free operations.
</p>
<div class="pullquote">
<p>
 Automating airplane control is, in many ways, easier than it is for cars.
</p>
</div>
<p>
 Moore likens these autonomy systems to riding a horse. The rider directs and supervises the horse, but the horse is sufficiently intelligent and competent “to place one foot in front of the other, stay on the path and not jump off a cliff.” The researchers hope to piggy-back onto fast-improving self-driving car technology. “DEP plus autonomy could create a whole new market for general aviation aircraft,” he says.
</p>
<p>
 His NASA colleague, Ken Goodrich, who is a senior research engineer at Langley and an expert on aircraft controls and automation, is somewhat less sanguine. “Human error is responsible for 90% of accidents both on the road and in the air, but nobody’s thinking about the new errors all this automation could produce,” Goodrich says. “So far, no one knows how to program common sense into machines.” Also needed is a new procedure that the U.S. Federal Aviation Administration could use to certify the safety of autonomy technology, he says.
</p>
<p>
 Goodrich does acknowledge, however, that automating airplane control is, in many ways, easier than it is for cars. “The road is a very cluttered environment, and you operate much closer to other vehicles, so the necessary reaction times are much shorter.” In contrast, “the sky is wide open” with aircraft spaced much farther apart. And with the automotive industry and companies like Google devoting huge resources to develop self-driving cars, autonomy may happen much faster than many industry observers imagine.
</p>
<p>
 But even without autonomy, DEP would be beneficial because electric motors are “a scale-free technology,” Moore says. “Current propulsion engines just don’t scale well. A full-size turbine engine can be around 40% energy efficient, but if you take it down to 100 horsepower, it’s only 24% efficient—6 horsepower per pound vs. 0.5 horsepower per pound.”
</p>
<p>
 Unlike fossil-fuel turbines, electric motors can be compact, reliable, and efficient. “They provide extremely good power-to-weight ratios—two times better than turbine engines and three times better than any reciprocating engine,” he says. And not only is “electric propulsion happy to scale to any size, you can place them anywhere you want.” Clever placement of electric motors could, for example, improve a plane’s lift and control.
</p>
<p>
 Today’s tiny, all-electric and hybrid-electric aircraft use electric motors that deliver several kilowatts of power, but they suffer from severely limited range because of battery constraints. To travel longer distances, next-generation battery-electric and hybrid-electric light aircraft and regional airliners that carry up to 50 passengers will need to produce, say, from 1 to 2 MW of power. Larger airliners will require 2- to 5-MW-class electric motors and power systems. While batteries have improved lately, they are still many years away from providing that sort of power at the light weight required for flight.
</p>
<p>
 Hypercommuter Planes
</p>
<p>
 That hasn’t tempered everyone’s enthusiasm, though. JoeBen Bevirt, serial entrepreneur and founder of Santa Cruz-based Joby Aviation, is one of e-aircraft’s true believers: “Six or seven years ago, everybody in the aviation industry thought we were completely nuts to be working on electric aircraft. Now that major companies like Airbus and Siemens have jumped into the field, everybody is taking notice.”
</p>
<p>
 Bevirt hopes to build two-seat hypercommuter planes, personal air vehicles that resemble George Jetson’s flying cars. These pioneering personal aircraft would produce zero in-flight carbon dioxide emissions and generate less noise while cutting overall operating costs by perhaps as much as a third. “For me, this technology is a chance to make aviation much more relevant to people’s day-to-day lives,” Bevirt says.
</p>
<p>
 Hypercommuters, which would be driven by arrays of small propellers, could initially operate as Uber-like “sky taxis” to speed the long-distance daily commutes of high-income workers who live in traffic-congested regions such as Silicon Valley. Such quieter, short-takeoff-and-landing aircraft could fly from a network of small public and private landing pads dotting urban areas.
</p>
<p>
 The hypercommuter concept’s small-diameter propellers are optimized for low rotational speeds, which means they produce less noise than larger props. In addition, each propeller can rotate at slightly different velocities to spread out the sound frequencies they emit, cutting “community noise,” researchers hope, by as much as 15 decibels. That means that a hypercommuter-type airplane could produce “100 times lower noise levels than those of helicopters,” Moore says.
</p>
<p>
 The multiple propellers also enhance safety, he says. “The safety statistics for general aviation aircraft are not all that great,” with most accidents happening during takeoff and landing—“when planes are flying low and slow.” Not only could DEP provide redundancy against propeller failure, it could maximize control. “With the blown wing, you have incredible lateral control. If one wing loses lift and stalls at low speed, causing the plane to roll off to one side, you can just power out of it.”
</p>
<h2>
 Larger Electric Aircraft
</h2>
<p>
 By 2018, the NASA team plans to apply DEP to “thin-haul” aircraft such as the nine-passenger, twin-engine Cessna 402 that airlines such as Cape Air in Massachusetts use for trips of less than 250 miles. “We think that DEP technology can deliver a 20–30% reduction in Cape Air’s total operating costs with zero in-flight emissions—the best of both worlds,” Moore says.
</p>
<p>
 Moore and his colleagues think electric aircraft have potential at even larger scales. “We realized that distributed propulsion is applicable to larger aircraft as well, even commercial transports flying stage lengths of around 600 miles. It could therefore also be a game-changer for turboprop and regional jets that the airlines fly today.”
</p>
<p>
 Should smaller electric airplanes start to gain traction during the coming decade—and battery tech keep pace—aerospace engineers could then begin developing larger aircraft with advanced DEP systems. Banks of electric motors along the wings could boost propulsive efficiencies by as much as 10% by sucking in the slow-moving boundary layer of air nearest the wing, NASA’s Del Rosario says.
</p>
<p>
 That would help save the energy compared with traditional engines, which hang below the wing and generate their own drag. The smaller propellers would be immersed in the slower-moving boundary-layer that flows over the wing, accelerating the already moving air and not generating additional drag. Boosting efficiency by 10% would be a major advance for the airline industry in particular, which today fights for every fraction of a percentage point to lessen fuel costs.
</p>
<p>
 The question is whether the fan can withstand such a turbulent environment without its efficiency being compromised. Early results say it’s possible. “First, can you build a fan that can withstand and operate in that distorted air flow?” Del Rosario asks. “Second, by ingesting the boundary layer, are we indeed cutting drag? Our latest research says yes.”
</p>
<p>
 As the tests on NASA’s HEIST big-rig and other projects move forward, it’s looking more and more like electric aircraft will become an everyday fixture at airports, Moore says. “We think that electric motors and DEP technology will become available on airplanes a lot quicker than most people think. It’s going to fundamentally change what we can do and what we’ve wanted to do for so long in aviation.”
</p>
<div class="newsletter"></div>
<p>
 Photo credits:
 <a href="http://projecthabu.com/post/119489255031/nasa-armstrong-flight-research-center-is">
  Curt Mason/Project Habu
 </a>
 , NASA, Joby Aviation
</p>

<p>Electric motors promise to revolutionize aviation, slashing carbon emissions while improving safety and maneuverability.</p>

A prototype wing equipped with plasma actuators.

Flying with the Fourth State of Matter

<p>
 A scaled-down gale blows over a flat plate set inside the tabletop wind tunnel. Despite the low lighting and hazy Plexiglas view portals, we can clearly see the frenzied fluttering of streamer ribbons, called telltales, in the field of little wind vanes that carpets the exposed test surface inside.
</p>
<p>
 At first, each unruly telltale flies every which way, clear evidence of unsteady air flows gusting within. “OK, that’s off,” the researcher says.
</p>
<p>
 “Here’s on…” Almost like a sorcerer’s spell, an otherworldly, blue-violet halo emerges at the front of the plate and hovers corona-like, casting peculiar purple shadows onto the walls. The telltales meanwhile become suddenly and strangely obedient, instantly swinging round in near unison, aligned by an insistent new wind.
</p>
<p>
 “Off,” she says. The ribbons flap arbitrarily as the eerie electric flame fades. “On.” More purple haze and parallel ribbons.
</p>
<p>
 “Off, off…and on.” The odd glow, curious order, and incessant roar of the fan drop away. By the time the lights come on, a whiff of ozone hangs in the air and everyone in the room is grinning uncontrollably.
</p>
<div class="related"></div>
<p>
 And for good reason. We just watched moving air being controlled by plasma, the lesser-known, fourth state of matter which also exists in the blistering core of our sun. And while such lab demonstrations are both uncanny and awe-inspiring, these so-called plasma actuators could produce far more impressive benefits in the real world, especially for the aviation and wind power industries, and maybe even the trucking business.
</p>
<p>
 On airplane wings, for example, tiny plasma actuators could help planes fly more safely, more efficiently, and with greater stability and control. They can speed, slow or divert air flows in ways that can cut drag, fuel use, and CO 2 emissions by as much as 25%, researchers estimate. Some experts even think that these devices might someday replace conventional flight control surfaces such as flaps and ailerons. Imagine witnessing the ghoulish purple glow of the lab demo from the window seat of a transcontinental flight.
</p>
<p>
 More immediately, aerodynamicists are looking to place the same technology on the huge, vulnerable, and costly blades of wind turbines to improve their efficiency, extend their lifetimes, and even help them more effectively cope with gusting winds.
</p>
<p>
 Surprisingly, this unearthly technology is, in fact, most commonly found hiding inside people’s homes.
</p>
<h2>
 A Plasma Wind
</h2>
<p>
 One of the more common uses of plasma wind devices is in the glowing blue-purple-violet “fan” elements of household ionic air purifiers. Electric fields emitted by surface electrodes cause the adjacent air to become energized, allowing its component molecules to break down into ions and electrons to form plasmas—glowing clouds of electrons and charged air molecules, mostly nitrogen and oxygen ions. A natural version of this “cold plasma” phenomenon is sometimes seen ghosting off ships’ masts and superstructures, and aircraft wings. There, it’s known as the legendary St. Elmo’s fire.
</p>
<p>
 Embedding high-voltage versions of these wind generators—flat, thin-film electrodes—in an airplane’s wings can not only summon artificial St. Elmo’s fire instantly and on command, but it can also send the charged plasma clouds—and air they carry along with them—moving in one direction or another. That is, these plasma actuators can readily accelerate, or “blow,” air where and when it’s needed with no moving parts, boosting lift or cutting drag depending on where and how they’re used.
</p>
<p>
 The field of plasma aerodynamics itself has an intriguing history. After some early conceptual work for spacecraft reentry in the 1950s to the 1970s, the field reawakened when word arrived in the West that toward the end of the Cold War Soviet researchers had incorporated plasma-based technology into their scramjet-powered Ayaks hypersonic vehicle concept to enhance combustion and aerodynamic performance. Russian engineers later detailed in a magazine article “how plasmas could weaken the shock waves generated during hypersonic flight,” says Jon Poggie of Purdue University, who at the U.S. Air Force Research Laboratory Command in Ohio previously studied plasma control of supersonic flows.
</p>
<div class="ad"></div>
<p>
 Some people called it “plasma magic,” but western researchers wanted to know more as speculation grew and initial experimentation began, he says. “If plasmas really could reduce drag, it was thought that it might be possible to delay the onset of sonic booms, steer aircraft by applying plasmas selectively to different parts of the vehicle, and even reduce heating around hot spots on an airframe.” In time, the Soviet Union collapsed and émigré Russian researchers arrived in the west, spurring fruitful research on plasma aerodynamics among R&amp;D institutions in Europe, the U.S., and Russia, an extraordinary international collaboration that lasted almost 20 years.
</p>
<div class="pullquote">
<p>
 “A tiny push at the right place and time can excite a much larger, and often positive, result.”
</p>
</div>
<p>
 Some people called it “plasma magic,” but western researchers wanted to know more as speculation grew and initial experimentation began, he says. “If plasmas really could reduce drag, it was thought that it might be possible to delay the onset of sonic booms, steer aircraft by applying plasmas selectively to different parts of the vehicle, and even reduce heating around hot spots on an airframe.” In time, the Soviet Union collapsed and émigré Russian researchers arrived in the west, spurring fruitful research on plasma aerodynamics among R&amp;D institutions in Europe, the U.S., and Russia, an extraordinary international collaboration that lasted almost 20 years.
</p>
<p>
 Plasma wind is created by applying an alternating or pulsed current to electrodes to produce “a dynamically changing electric field,” says Steve Wilkinson, an aerodynamicist at NASA Langley, one of a cadre of researchers at government, industry and university labs in the U.S., Europe, Russia and elsewhere who investigate what’s called plasma aerodynamics. The applied voltage difference causes the charged ions in the plasma to shift and “some ions collide with nearby air molecules, which rebound in the same direction, creating wind,” he says.
</p>
<p>
 “But these are not thrusters,” says Thomas Corke, a leading plasma aerodynamicist at the University of Notre Dame. “They blow like weak table fans, only enough to move notebook pages around.” But, paradoxically, if aimed and timed correctly, the breezes blown by these plasma actuator devices can be strong enough to significantly alter how wind flows over wings in beneficial ways, he adds.
</p>
<p>
 “With growing computer power and practice, we can now model the detailed behavior of the airflows over the wings, the fluid dynamics, as well as behavior of the plasmas themselves with increasing precision,” Corke says. “This lets us identify natural instabilities in air flows that we can exploit with great effect because just a tiny push at the right place and time can excite a much larger, and often positive, result.”
</p>
<h2>
 Plasma Controls
</h2>
<p>
 These subtle nudges could radically change the way airplanes are designed and flown. “This technology can provide the leverage that airplane designers need to implement a range of potentially significant performance improvements,” Poggie says. Small interventions in the airflows over wings and tail surfaces, he says, could reduce drag, augment lift, enhance efficiency, muffle noise and vibration, or boost flight control.
</p>
<p>
 Conceptually, Poggie says, researchers envision tiling wings and tail surfaces with arrays of flat plasma actuators that could be individually activated to tweak airflow to improve aerodynamics. The wall-to-wall electrodes “could even heat up for deicing in cold weather.”
</p>
<p>
 Theoretically, plasma actuators can be effective at all speed ranges, from slow drones to fast jets and even hypersonic vehicles, but so far the technology has been tested mostly under ideal conditions in labs, wind tunnels, or on model aircraft. But as long as fluid mechanicians and aerospace engineers have correctly simulated the plasma behavior and the airflow over an airplane in most flight conditions—no straightforward task—an effective strategy could be developed to fine-tune the flows. “You could dial-in and modulate the flows, tuning them to do what you need to do,” Poggie says.
</p>
<p>
 The power of such a fundamental approach was highlighted by a recent NASA-funded demonstration project in which Corke’s team at Notre Dame succeeded in deploying plasma actuators to suppress what he describes as “an important near-surface flow instability” that aerodynamicists know presages the onset of deleterious turbulence and drag.
</p>
<div class="pullquote">
<p>
 “Plasma actuators could extend a Predator UAV’s range by 300%.”
</p>
</div>
<p>
 Corke and his team reported that the wind tunnel test item, which used “new actuators that developed 20 times more thrust while consuming 100 times less power, produced a 65% drag reduction.” The Notre Dame researchers have found that introducing a small oscillation whose waves move perpendicular to the air flow path can halt the onset of the so-called near-surface flow instabilities that lead to turbulence.
</p>
<p>
 The results, when scaled up, could be impressive. “I ran the numbers for an aircraft design the size of a Predator UAV on a computer,” Corke says. “If you put plasma actuators on one, it could extend its range by 300% and its endurance from 24 to 36 hours.” NASA has awarded the group a Phase 2 R&amp;D contract, and now DARPA is said to be interested in the novel technology, which has since been patented and licensed to a start-up company.
</p>
<p>
 In the near term, plasma actuators offer an attractive alternative to competing aviation flow-control technologies, says Rasool Erfani, a lecturer of mechanical engineering at the Manchester Metropolitan University in the U.K. Plasma-based devices have a very low profile and can provide endless, exact puffs over short time scales.
</p>
<p>
 The technology could even replace traditional wing flaps, he says. Although the plasma-actuated “flapless” or “hingeless” wings are currently only suitable for smaller aircraft such as drones or UAVs, Erfani believes that further development could make the technology viable on larger, faster planes.
</p>
<h2>
 More Subtle Adjustments
</h2>
<p>
 But that will take better plasma actuators. The most common systems, known as dielectric barrier discharge (DBD) devices, pair high-voltage electrodes separated by an insulator, typically a temperature-resistant polymer, glass, quartz, or ceramic. When activated ionized air spreads from the front electrode to the insulated rear one, creating a plasma trail. The rear insulator doesn’t readily conduct the current so it prevents arcing between the electrodes, which allows the electric fields to break the air down into plasmas.
</p>
<p>
 DBDs will, however, need to be more effective if they are to be more widely used, says Subrata Roy, a professor of mechanical and aerospace engineering at the University of Florida who with his colleagues has developed a new design that is more versatile than standard units. They use electrodes arranged in serpentine, or sine wave-like, patterns, rather than rectilinear rows. These new geometries can, for instance, blow in three directions at once or create three-dimensional aero effects such as tiny tornado-like vortices that can seed similar larger whirlpools in much larger airflows.
</p>
<figure class="richtext-image full-width">
<img alt="" class="richtext-image full-width" src="https://staging.pbs.org/wgbh/nova/media/images/wind-tunnel-oil-flow.width-990_4uMpefH.jpg"/>
<figcaption>Airflow over a traditional wing is measured in a wind tunnel using fluorescent oil.
        
    </figcaption>
</figure>
<p>
 Roy’s team is also looking at using new materials, such as insulators made of aerogels, the lowest density, lightest material known. Other plasma actuator research focuses on improved pulsing, phasing, and timing of the electric currents and the fields they create.
</p>
<p>
 The other big challenge for flight applications are the weighty power supplies for the high-voltage current. “The trouble is that making plasmas is not at all energy efficient,” NASA’s Wilkinson says. “It takes a lot of energy to dissociate the air molecules and move them about.” But progress elsewhere toward
 <a href="http://www.pbs.org/wgbh/nova/next/space/electric-aviation/">
  electric aircraft
 </a>
 —equipped with hybrid systems and better batteries—could help overcome that limitation since electric actuators are a natural fit in electric aircraft.
</p>
<p>
 The fact that DBDs and similar plasma actuators are relatively simple means that the technology poses a relatively low barrier to entry, which helps account for the continued popularity of plasma aerodynamics research. It’s the kind of thing that a graduate student can set up and demo, several researchers say, though truly understanding the devilishly complex air and plasma flows is another thing entirely.
</p>
<p>
 Engineers are also looking to apply plasmas elsewhere on airplanes. GE, for example, is investigating using the technology in turbine engines: a first-ever full-scale compressor application has been demonstrated. GE researchers are also exploring the use of a related cold plasma technique in lean-burn combustors for aircraft engines to stabilize the flame and keep it lit as fuel input is reduced. Another potential application is to break down the fuel-air mix into ions to make it burn more readily in engines of all kinds.
</p>
<h2>
 Back to Earth
</h2>
<p>
 With Russia and the West now again at odds, the previously successful aerospace research collaborations have ended. Government support has ebbed, leaving many practitioners scratching for funding. It is perhaps unsurprising that the most promising near-term application for the technology is wind power, which has no problem with heavy power supplies and offers the possibility of a more immediate return on investment.
</p>
<p>
 “Wind turbines could greatly benefit from plasma DBDs,” Purdue’s Poggie says. Turbine blades must cope with heavy gusts and constantly varying, unsteady airflows that rob efficiency and stress the composite blades over time, he says. Existing mechanical fixes such as feathering, or rotating, the blades to less-exposed angles have proved either problematic or unsatisfying depending on what they’re attempting to address.
</p>
<p>
 “The concept has been around for awhile to tap off some of that generator power to run plasma actuators on the blades,” Poggie says. “That would allow the blade to respond to changes in wind speed and direction much faster to better maintain maximum power production.”
</p>
<div class="newsletter"></div>
<p>
 “Wind turbine blades constantly bend, bend, bend—and are built to withstand it,” he says. But over years of constant stress and strain, these high-cost components eventually fail from mechanical fatigue. Plasmas actuators could significantly reduce those loads to extend blade lifetimes considerably, saving wind farm operators money.
</p>
<p>
 Another terrestrial technology that could benefit from plasma aerodynamics are big-rig trucks and other highway vehicles. Plasma DBDs could be installed on the surfaces of truck bodies where they could smooth turbulent airflows and cut drag and thus fuel consumption rates, particularly at higher speeds. Plasma Stream Technologies, a small Iowa firm that’s developing the concept, recently licensed the technology and design for a power supply from Notre Dame based on past research done with GM and NASA, according to news reports.
</p>
<p>
 In time, it’s possible that expertise developed closer to Earth will allow engineers to wield plasma in the skies, where their research began. But whether it’s airplanes, wind turbines, or trucks, the promising efforts to master St. Elmo’s fire will likely someday bring an odd, new glow into the world.
</p>
<p>
 Photo credits: Manchester Metropolitan University, NASA
</p>

<p>Some experts even think that these devices might someday replace conventional flight control surfaces. More immediately, aerodynamicists are looking to place the same technology on the huge, vulnerable, and costly blades of wind turbines.</p>

Using Artificial Intelligence to Spot Hospitals’ Silent Killer

<p>
 Barbara, a 52-year-old woman, appears one morning at a Baltimore hospital emergency room complaining of a sore foot. After the EMS physicians examine her, they diagnose dry gangrene of the toe and admit her to a general ward for monitoring.
</p>
<p>
 On the third day, Barbara shows signs of what looks like pneumonia. They inspect her and prescribe the usual treatment, a course of antibiotics. However, three days later, Barbara’s heart rate suddenly rises, and in the afternoon, she starts breathing faster. (Barbara is not her real name.)
</p>
<figure class="richtext-image full-width">
<img alt="hospital monitor" class="richtext-image full-width" src="https://staging.pbs.org/wgbh/nova/media/images/hospital-monitor.width-990_3zy0alk.jpg"/>
<figcaption>A patient's vital signs being monitored in the hospital.
        
    </figcaption>
</figure>
<p>
 During the following twelve hours, her condition deteriorates rapidly until she enters septic shock—a severe condition with mortality rates as high 50%. Barbara’s doctors hurriedly prepare her for transfer to the intensive care unit.
</p>
<p>
 But, unfortunately for Barbara, it is already too late. Despite receiving every possible treatment during her next three weeks in the ICU, the woman grows steadily worse until her kidneys then lungs finally fail, and on the 22 nd day of her hospitalization, she dies.
</p>
<div class="related"></div>
<p>
 Barbara’s mild foot infection had escalated into fatal septic shock nearly entirely undetected.
</p>
<p>
 Barbara’s story is far from unusual. Every year, about 250,000 people in the U.S. die from sepsis, which occurs when the body exhibits an extreme reaction to an infection. Diagnosis is particularly difficult because sepsis often shows no obvious signs until organ dysfunction has already begun, says Suchi Saria, a machine learning researcher at Johns Hopkins University whose current focus is precision health care. “Early interventions such as antibiotics have been shown to improve outcomes,” Saria says, “but the disease often goes undetected for too long.”
</p>
<p>
 Driven by the nation’s aging population, the rise of drug-resistant bacteria, and the growth of medical interventions, sepsis has become one of the most common reasons for hospitalization in the U.S. More than 1.5 million domestic sepsis cases now occur annually, which accounts for more than $20 billion a year in health-care spending, stemming mainly from the need for more intensive care and longer hospital stays.
</p>
<p>
 Worse, sepsis now contributes to up to half of all deaths of hospitalized patients, “making it the eleventh biggest cause of death in the U.S., killing more people than breast cancer, prostate cancer, and AIDS combined,” Saria says.
</p>
<figure class="richtext-image full-width">
<img alt="" class="richtext-image full-width" src="https://staging.pbs.org/wgbh/nova/media/images/blood-cultures.width-990_mZ1A8m9.jpg"/>
<figcaption>Tech. Sgt. Elizabeth Ehrnschwender unloads blood cultures being monitored for the presence or absence of microorganisms in blood.
        
    </figcaption>
</figure>
<p>
 She says she tells Barbara’s story because it helps cut through the scary statistics to personalize the ongoing tragedy of sepsis in America. But it’s also because she and her team at the Hopkins’ Machine Augmented Decision Making Lab think that Barbara is just the kind of patient that their new computer system should be able to help save, perhaps soon.
</p>
<p>
 “The signs that help pinpoint the diagnosis may already be in your data, but the data in electronic medical records are really messy,” Saria says. “We’re designing computer algorithms based on statistical and computational techniques that were recently developed to allow clinical experts to identify sepsis faster.”
</p>
<div class="ad"></div>
<h2>
 When Delay Means Death
</h2>
<p>
 Researchers have shown that every hour a sepsis victim goes untreated adds about 8% to their mortality rate, says Dr. Peter J. Pronovost, who directs Johns Hopkins Medicine’s Armstrong Institute for Patient Safety and Quality and is collaborating with Saria on this work. “So, every second’s delay in recognizing it is critical—literally a matter of life or death.”
</p>
<p>
 “The key with sepsis is early diagnosis,” says Dr. Emanuel Rivers, a professor of critical care medicine at Henry Ford Hospital in Detroit and a pioneer in the development of the procedures widely used today to detect and treat the condition. “You need to closely monitor high-risk patients, to recognize any signs of sepsis as soon as possible, and then have something to do about it that works.”
</p>
<p>
 Since Rivers and his colleagues published that sepsis work in 2001, “what had been ignored has become a hot area,” he says. “A whole group of scientists have focused on early diagnosis, bringing to bear different expertise, different technologies and approaches.”
</p>
<div class="pullquote">
<p>
 “The key with sepsis is early diagnosis.”
</p>
</div>
<p>
 To improve sepsis care, Rivers continues, many health research organizations have begun to implement various screening tools that can send an alert or alarm to a clinician or nurse when a patient meets certain criteria that classify them as high risk. He says that at least five healthcare companies are planning to introduce commercial sepsis-alert products soon, noting that they resemble existing monitoring/alert products for cardiology patients. “A computer algorithm analyzes all the health signals, such as vital signs and metabolic levels, from the patients’ electronic records to tell which have the greatest likelihood of getting sepsis.”
</p>
<p>
 “The trick is to monitor all the variables that really mean something,” Rivers says, “and to, as they say, ‘get rid of the noise in the data’ that could lead to system hiccups where the monitoring data somehow gets translated into some unwanted action.”
</p>
<h2>
 Crying Wolf
</h2>
<p>
 “One key issue with these screening systems is false alarms,” Pronovost says. “You can’t have the alarm system crying wolf or it will be ignored.” So-called alarm fatigue results when people learn to ignore computerized flags, alerts, and notifications because too many turn out to be false.
</p>
<p>
 In 2015, Saria and her team first showed that a computer algorithm they developed could sift through patients’ records and predict septic shock—the deadliest version of sepsis—in 85% of cases, usually more than a day before onset. Two-thirds of the time the system predicted sepsis before it inflicted any damage. That’s 60% better detection performance compared to current screening tests without raising the rate of false alerts.
</p>
<p>
 Saria believes that reducing the rate of false alerts further is crucial for practical adoption. “Doctors are busy people, and their every moment is valuable.”
</p>
<h2>
 The Bayesian Way
</h2>
<p>
 Medical data is challenging to work with because of widespread omissions, inputs from various types of instrumentation with different measurement practices, and other dissimilarities that had always made such computer analysis seemingly intractable. The team’s tool, Targeted Real-time Early Warning System, or TREWS, aims to deploy these machine-learning methods inside hospitals to help save lives.
</p>
<p>
 The researchers realized early on that AI algorithms that leverage so-called Bayesian techniques in probability and statistics might do the trick because “they are fundamentally suited to dealing with uncertainty,” Saria says.
</p>
<p>
 Bayesian machine learning allows researchers to encode in models their prior beliefs about what those models should look like and how they should behave. Then, as additional information comes in, they can update those beliefs. To get a feel for how Bayesian inference works, consider the example of a physician who sees a patient complaining of chest pain. The doctor will have a very different expectation of whether the diagnosis is heart disease if the patient is a 65-year-old man versus a 12-year-old boy. As she receives new evidence such as electrocardiogram (ECG) test results, the doctor will update her beliefs to incorporate the new information. Bayesian algorithms can similarly adjust to new information, but at the same time are able to operate in domains where data is patchy and sometimes inaccurate.
</p>
<figure class="richtext-image full-width">
<img alt="" class="richtext-image full-width" src="https://staging.pbs.org/wgbh/nova/media/images/e-coli-ndm-1-cropped.width-990_4z1kEi1.jpg"/>
<figcaption>
</figcaption>
</figure>
<p>
 This flexibility is critical for models that are used in healthcare because “medical data comprise hundreds of different signals each with its own significance and effect—or not,” Saria says. In fact, it’s a moving haystack, hiding a few, potentially fake needles. While there are many signals, not all of them are relevant to forecasting for a specific patient, plus many readings are missing at various points, and data errors are often frequent. The probabilistic foundations for Bayesian machine-learning techniques make it well equipped for characterizing the uncertainties that accompany flawed data sets, allowing better, cleaner evaluation of the quality of the modeling results.
</p>
<p>
 “It’s a question of signal-to-noise, of competently estimating and always propagating the uncertainties associated with the entire system and each part,” Saria says. Essentially, each data point is always linked with a certain uncertainty, or error bar, that serves as a gauge of the signal’s trustworthiness. That uncertainty can be incorporated into the future modeling analyses, ensuring that it’s not lost along the way. Knowing the uncertainties associated with each data point ultimately makes the final result that much more valuable.
</p>
<h2>
 Embracing Uncertainties
</h2>
<p>
 TREWS uses Bayesian inference methods to propagate the uncertainties of each medical data point throughout the analysis. For example, this approach is able to start out by asking—and then accommodating—something as basic as, how accurate and reliable are all those blood pressure measurements in the first place? It automatically weights more accurate or trustworthy variables more heavily in the output, and it reduces the weight given to less reliable measures—exactly the sort of strategy that well-trained humans use.
</p>
<p>
 This machine learning approach differs from the more familiar AI technique of deep-learning, which are interconnected stacks of mathematical optimization functions that can operate in parallel and “learn” by tackling similar tasks or problems over and over again. The latter have shown tremendous success in diverse imaging-based, pattern-recognition, gaming, and other applications.
</p>
<div class="newsletter"></div>
<p>
 “But in medicine, where reliability is paramount,” Saria says, “we need techniques that are well-adapted to solving time-based prediction tasks with messy, limited data.” In joint work with Hossein Soleimani, a postdoctoral fellow in Saria’s lab, they show that their newer Bayesian way boosts “true positive” rates of detection by 200% to 300% over alternate state-of-the-art approaches, she says.
</p>
<p>
 “What is also novel about this system that is that it could recommend the best personalized course of treatment for each sepsis patient,” Pronovost says. When the system recognizes that the patient is at risk, it gives care providers the option to tailor their therapy based upon the type of infection they have.
</p>
<p>
 Rivers, the sepsis expert, thinks that improved, TREWS-type analysis of patient monitoring could in time “turn up some new clues about sepsis” that lead to more effective therapies. He suspects, for instance, that close monitoring of patients’ biorhythms could provide useful clues that would justify triggering alarms at certain points.
</p>
<h2>
 Real-World Trials
</h2>
<p>
 Introducing a new predictive tool into the complex, often-interrupted and high-stakes environment of a hospital is challenging, something that the TREWS team is now focusing on as it prepares for real-world trials at Howard County General Hospital in Columbia, Maryland. These trials will determine whether the interactive system actually works in the field. The prototype algorithm is being embedded into the hospital’s electronic medical records system.
</p>
<p>
 There’s still lots to be worked out, from physician and nurse workflows, general usability tweaks, data validation, and training. They all have to be addressed beforehand if the implementation is to be successful. Ultimately, Saria says, TREWS is designed to seamlessly integrate into doctors’ routines, saving them time in sifting through the information they need to make decisions.
</p>
<figure class="richtext-image full-width">
<img alt="" class="richtext-image full-width" src="https://staging.pbs.org/wgbh/nova/media/images/enterobacteriaceae_6DhXKjH.width-990_2Vl0ESd.jpg"/>
<figcaption>
</figcaption>
</figure>
<p>
 “We need the systems to produce answers that can be relied upon,” she says. “We designed the system to operate in real-time and be highly interactive. But to start, we conducted a pilot trial in the background that allows us to measure physicians’ behavior and how the system could affect their practice.”
</p>
<p>
 “Right now we’re digging into where it should live within the clinician’s workflow,” says Dr. Anirudh Sridharan, a hospitalist and geriatrician at Howard County General Hospital who is working with Saria’s team. According Sridharan, the researchers are currently homing in on the best way to notify clinicians of potential sepsis cases and develop effective protocols for responding.
</p>
<p>
 “We’re addressing things as basic as ensuring we’re alerting the right person on the team,” Sridharan says. “Asking questions like: Should the alerts be active? Passive? Do you want a text notification on your phone? It’s all part of deploying the tool to best effect.”
</p>
<p>
 “We’re deploying a tool that can potentially deliver orders of magnitude of better care by surfacing the affected patients earlier so we can help them before it’s too late,” he adds. “And from what I’ve seen, we’re going to see a lot more AI in hospitals and health care.”
</p>
<h2>
 Making Use of Electronic Medical Records
</h2>
<p>
 It’s been a long wait for electronic medical data technology to truly pay off, Sridharan points out. “The old care model basically has been to replicate what was on paper records, and it certainly brought advantages such as better, updatable, trackable records that, for example, eliminated mistakes from chicken-scratch scripts and so forth,” he says. “But what everyone’s been waiting for is for everybody to stop just entering digital data, and somebody to finally take all this digital information and turn it into a real aid in making medical decisions.” Earlier systems provided “answers that were so simplistic as to be useless,” he adds.
</p>
<p>
 “Electronic medical records are the single greatest, most disagreeable, disappointment of modern practice,” says Dr. Michael Gropper, chairman of critical care medicine at the University of California, San Francisco’s School of Medicine. “We haven’t been smart about how we created the electronic records, and now we’re paying the price.”
</p>
<p>
 Meanwhile, the practice of medicine has changed, Gropper says. “Medicine used to be based on the hero model with the physician in the role of single, all-knowing decision-maker. That’s all changed; medicine is a team effort now.”
</p>
<p>
 “Improved outcomes rely on leveraging all the members as well as all the technology to deliver the very best care. But when we’ve introduced these kinds of screening tools into our ICU, we’ve found that the transition is usually not straightforward,” he warns.
</p>
<h2>
 Changing the System
</h2>
<p>
 Dr. Jonathan H. Chen, an expert at medical informatics at Stanford University, agrees. “Trying to unlock the power of electronic medical data sources to improve clinical care outcomes is a great goal, but I believe implementation will be harder than some may expect,” he says. “First, the high accuracy measures that researchers report may just be predicting things the doctors already know or are false alarms that are only distracting them, leading to alert fatigue.” Second, “just because an accurate prediction is made, doesn’t mean we have an intervention to change the outcomes.”
</p>
<p>
 “For example, the TREWS model was trained on data from patients already in the ICU receiving the highest level of monitoring,” Chen points out. “But having an ‘early warning system’ doesn’t make much sense at this point.” In addition, TREWS appears to be predicting sepsis in cases where doctors had already recognized the signs and were intervening. “The number you care about is positive predictive value,” he says. “Given that the alarm goes off, what is the chance the patient will end up having septic shock?” After a look at the TREWS papers, he calculates that roughly two out of three times the algorithm would be giving doctors a false alarm.
</p>
<p>
 “Of course, it’s easy to point out these technical issues,” Chen says, “because I commend Professor Saria’s team for continuing to push through and actually get systems implemented and evaluated in real clinical settings. This takes an enormous amount of work, resources, and political will to accomplish, far more than the relatively sterile steps of data analytics to produce the prediction model.”
</p>
<p>
 Just how Saria’s AI agent has gotten good enough to even try to wade through mountains of heterogeneous hospital data to sniff out something as subterranean as sepsis has a lot do to with her frustration with the so-far failed promise of electronic record-keeping. Like the other researchers, she longs for progress. “For seven or eight years now, we’ve had the digitized medical record data, but not much has happened yet to really exploit it,” she says. “I want to change that.”
</p>
<p>
 Photo credits:
 <a href="https://www.flickr.com/photos/yannconz/15785368136/">
  Yann/Flickr (CC BY-NC-SA)
 </a>
 , U.S. Air Force,
 <a href="https://www.flickr.com/photos/nathanreading/6721556041/">
  Nathan Reading/Flickr (CC BY-NC-SA)
 </a>
 , CDC
</p>

<p>Sepsis contribues to as many as 50% of hospital deaths. But a new tool could help doctors spot it before it&#x27;s too late.</p>

Steven Ashley

Articles

Synthetic Food: Better Cooking Through Chemistry

Flying the Electric Skies

Flying with the Fourth State of Matter

Using Artificial Intelligence to Spot Hospitals’ Silent Killer

Support Provided By