In This Issue
Engineering and the Health Care Delivery System
March 1, 2008 Volume 38 Issue 1
Volume 38, Number 1, Spring 2008. There are abundant reasons for the problems in health care delivery. Engineers may not be able to solve all of them, but the benefits of working toward solutions can be tremendous, and the challenges they present are enormously intellectually stimulating.

The Convergence of Information, Biology, and Business: Creating an Adaptive Health Care System

Saturday, March 1, 2008

Author: Christopher Meyer

As biology becomes an information science, health care will learn from nature how to accelerate change.

The world of health care lies between the realms of medical science, which continues to deliver new treatments and information at an accelerating rate, and of policy, which is conservative by design to protect patients’ safety. Other organizations and businesses have been adapting to changing environments much more quickly than health care, but as biology and information sciences converge, the health care system could use similar techniques to improve the quality of care.

Adaptive Enterprises
Since the days of Henry Ford, traditional businesses have prized stability, efficiency, and predictability. A traditional businessperson looking at the two graphs in Figure 1 might prefer the one on the left because of its stability and narrow range of variation. However, these graphs depict the heart rates of two individuals. The person on the right is healthy; the person on the left died eight days later. One might think a healthy heart would beat at a stable, predictable rate and that a diseased heart would have an erratic and changeable heartbeat. In fact, a healthy heart takes inputs from the body about the concentration of oxygen and sugar in the bloodstream and other conditions and then adapts—that is, it “makes a decision” about what to do. A heartbeat as constant as a metronome is a sign of disease.

Two cardiograms
FIGURE 1 - Two cardiograms showing (left) an unhealthy heartbeat and (right) a healthy heartbeat.

This analogy holds true for business. In 2006, Ford had to close a plant near Atlanta that had produced the company’s line of Ford Taurus and Mercury Sable sedans, essentially identical cars. Although the plant was one of the most productive on the continent, it was not adaptable.

In the same year, Honda constructed a plant that could produce both Civics and Elements, two quite different models. The capital efficiency lost from investing in adaptability was small compared with the robustness gained from being able to follow the market. Asian manufacturers like Honda understand that betting on correctly anticipating the market in an environment of constant change is bound to be expensive in the long run.

But these trade-offs are becoming increasingly old-fashioned, because robustness and efficiency are not necessarily mutually exclusive. Information-intensive approaches, such as mass customization and digital optimization and control, can eliminate the inefficiencies of variation, and real-time feedback loops and learning systems can update the way businesses work as circumstances change. In some industries, however, particularly health care, management approaches have not kept pace.

Learning from Nature
In the words of Stuart Kauffman (1993), a physician and complexity theorist, “Nature’s been [adapting] for 3.8 billion years. Maybe we should pay attention to how it does [this].” In the biosphere, adaptation refers to processes by which populations of organisms evolve in response to changing environmental pressures. How does nature do it, and how can managers adopt nature’s techniques?

In It’s Alive, Stan Davis and I created a simplified model of evolution as the outcome of six principles and then translated those principles into business terms (Meyer and Davis, 2003):

  • Self-organization. Any group of “decision-making units” capable of linking to each other, be they molecules, software routines, or people, is likely to organize autonomously to create something more complex. In nature, 92 chemical elements organized themselves into our planet. In business, farmers have been forming weekly village markets for centuries (Kauffman, 1993).
  • Recombination. Biology remixes the recipes for species by combining the genetic recipes of parents. This remixing, or recombining, is the source of almost all innovation in nature. In business, the Wright brothers created something completely new by combining the capabilities of the airfoil, the bicycle wheel, and the internal combustion engine.
  • Selective Pressure. The environment determines the likelihood of passing along genes to the next generation. A specific innovation, whether in the rain forest or the marketplace, may or may not confer an advantage. However, in a rapidly changing environment, a higher rate of innovation improves the odds of a species finding a successful adaptation.
  • Adaptation. Over time, with cycles of recombination and selection of the fittest, the abilities of a species evolve.
  • Coevolution. Species share the environment. As Kauffman says, “When the frog evolves a sticky tongue, flies get Teflon feet.”
  • Emergence. From looking at the periodic table, one would not predict that life would emerge from these diverse elements. However, interactions among self-organization, recombination, selection, adaptation, and coevolution lead to a connected ecology, or, in business, an economy made up of markets. Inter-dependencies among interactions make it impossible to predict which outcome will emerge. For instance, we cannot predict how a rain forest will change as the climate changes, or how capitalism will evolve as energy becomes scarce.

“General Evolution”:
Applying Biology to Business
Adaptation is embedded in biological systems, but the general theory can be applied to other kinds of complex systems, including social systems like business. When companies interact with one another they create new economic opportunities and capabilities. Business capabilities are like genes, and we can think of the economy as a set of capabilities turning each other on and off the same way a genetic regulatory network turns gene expression on and off.

For a business to thrive in a volatile, evolving environment, it must apply the lessons learned from biological evolution. Managers must shift from predicting and controlling change to building an organization that senses external change and responds appropriately, like a healthy heart. We refer to this as the shift to “adaptive management.”

The six principles of evolution outlined above can be translated into six practical business principles, which we call “memes for managing” (Table 1). In the following sections we describe the business application of two memes: (1) “recombine” and (2) “sense and respond, learn and adapt.”

TABLE 1 Memes for Managing


Name Definition Business Example Application to Health Care
Self-organize Manage your organization from the bottom up. Influence the rules that affect individual choices rather than the overall behavior of the organization. Southwest Airlines cargo-routing simulator Hippocratic oath,
rule-based HMOs
Recombine Proliferating connections make recombination—of software code, product attributes, people, and markets—easier. Turn your business into an open system to capture the value and innovation of diversity. Genetic algorithms to schedule John Deere factory production Gene shuffling
Sense and respond, learn and adapt Equip your business to sense changes in real time and to respond immediately, accurately, and appropriately. Then learn from that experience and incorporate the new information into your repertoire of responses. Honeywell Adaptive Intelligent Recovery Thermostat Partners HealthCare physician order-entry system
Seed, select, and amplify Test many diverse options and reinforce the winners. Experiment, don’t plan. Public beta software releases; Sun’s Java Fund Evidence-based medicine Public beta software releases; Sun’s Java Fund Evidence-based medicine
Live at the edge of chaos The rate of environmental change demands internal instability for survival. Disrupt the static elements in your organization. Google, Amazon Dr. Peter Pronovost’s checklists to prevent line infections in ICUs
Monetize molecules The next wave of innovation is in the intersection of molecular science and technology. Hybridized plants, nanotechnology Drug discovery

“Recombine” at John Deere
John Deere uses techniques of directed evolution—the intentional recombination and selection techniques used to create thoroughbreds and show dogs—to schedule operations in a highly complex factory in Moline, Illinois. The factory makes seed planters, the large machines farmers tow behind tractors to sow seeds.

Manufacturing planters is challenging because the many kinds of seeds, tractors, soils, and climate generate 1.6 million configurations, most of which are produced in low volumes. Thus Deere could not build a truly automated-transfer assembly-line factory. As a result, workers wheeled planters through the aisles manually, and, depending on the particular configurations of the planters, they often bumped into each other and impeded each other’s progress. Despite the use of standard optimization techniques, such as linear programming, throughput was poor.

Finally, Bill Fulkerson, a Deere engineer, proposed a nonlinear approach—genetic algorithms. Given the set of planters the factory wanted to build the next day, the system created a few random sequences for producing them. The schedules were expressed digitally—each one a string of zeros and ones, just as the instructions for creating an organism are expressed in a string of Gs, Ts, As, and Cs.

Deere engineers then applied the principle of selection to the schedules, evaluating each one with a computer simulator that assigned it a “fitness” score based on its simulated throughput. The winning sequences were then “put out to stud,” that is, a genetic algorithm recombined parts of the best schedules to create a new generation of schedules.

Every night, 40,000 new schedules ran simulated races, and the winning plan was used for the real-life production on the factory floor the next day. The genetically engineered schedules were about 15 percent more efficient than the schedules Deere had designed based on traditional linear methods.
The lesson from Deere’s recombinant strategy is that nature’s approach to adaptation can outperform previous engineering methods. By using genetic algorithms, Deere literally translated the recombination of DNA to recombinant code in another domain.

“Sense and Respond, Learn and Adapt” at Honeywell and BMW
The meme “sense and respond, learn and adapt” (SRLA) has two feedback loops, one short term and one longer term. The thermostat in your home has one of these loops, a real-time sense-and-respond feedback system that senses the temperature in the room and responds by turning on the furnace when it drops below a designated level.

By comparison, the Honeywell Adaptive Intelligent Recovery Thermostat has higher consciousness. It observes the effect of its actions on a room, and if it overshoots the mark one day, perhaps because of the high thermal mass of the radiators, it remembers and turns off the heat earlier the next day. That’s the “learn” part of SRLA.

As the seasons change, the effect of the learned behavior also changes, for example because the room loses heat faster in colder weather. Because the thermostat keeps learning, its behavior adapts to these seasonal shifts. It not only senses and responds, but also learns from the results of its actions and adapts to match its performance to the shifting environment.

Recombination and SRLA together can lead to a new phase of economic evolution enabled by the convergence of information and business. BMW recently ran an ad showing a car’s wheels screaming and spinning on an icy road. The driver hits a button and talks to the concierge, who quickly finds a solution to the driver’s dilemma. The BMW downloads a patch to its traction-control system, and off goes the car with new behavioral rules onboard.

Nature’s methods
can outperform
engineering methods.

However, the new code could have evolved without human intervention. The antilock brake system (ABS), which knows when a wheel is spinning, could have transferred the telemetry to a central information system. Even if the exact configuration of weather condition, climate, load, tire design, and road surface had never been seen before, BMW could have used a genetic algorithm to invent a new software solution by using a simulator to evaluate the problem (as at Deere) and send the “winner” to the traction-control system.

So far we have “sense” (the ABS system), “respond” (telemetry), and “learn” (recombination via a genetic algorithm and the installation of new code in the braking system). What about “adapt”? BMW could take note of the success of the new software and update the ABS not just in one BMW but throughout its fleet. In this way the “species” would evolve, or adapt, in response to the experiences of individual “organisms” to their environment.

Businesses that install and exploit these feedback loops, and have the capacity for innovation to adapt to them, can create the capacity to respond to changing environments, not through studies, surveys, and redesigns, but through a continuous evolutionary process.

Applying Biology to the Business of Health Care
In the words of Jerry Grossman, former head of the New England Medical Center, “Health care has been locked in irons since the end of World War II.” While most of the business world has become more adaptable, shortened its product life cycles, increased process innovations, and increased the rates of job and CEO turnover, health care, like the heart of the sick man in Figure 1, has been too stable for too long.

Keep in mind, however, that health care’s reluctance to embrace change is based on valid arguments. Experimentation—and innovation always begins with experimentation—can be dangerous. Like an industrial factory floor of the 1930s filled with dangerous technologies such as molten steel and explosive chemicals, hospital environments can be dangerous places, and mistakes can be fatal. Innovation is not a good thing if people die when it leads to an unexpected result. The medical meme for health care management, “First, Do No Harm,” is conservative for good reason.

Industry has increased its pace of change in the past two decades by embracing information technologies. Real-time sensing, algorithmic control systems, and simulation, for example, have expanded the performance envelope of all sorts of businesses.

Innovation begins with
experimentation, which can
be dangerous, especially in
health care.

Today, with the convergence of biology and information, health care is poised to follow suit, but the industry is still trying to operate in two places at once—a capital-intensive, dangerous, cost-imperative world and, simultaneously, the area of convergence of medical science and information technology, which is accelerating change and compelling adaptation. Crossing the divide will require a new kind of experimentation.

Medical researchers today can run experiments in silico, with computer simulations, rather than in vivo, through clinical trials. In this way they can begin to understand how complex biological systems recombine, sense and respond, and learn and adapt as a collective response. Scientists can simulate, for instance, a fully functioning human heart. Researchers at UCLA have studied computer models of a heart to determine how drugs could be targeted more directly to the turbulent breakup of heart waves during sudden cardiac arrest (Xie et al., 2004). The Food and Drug Administration now accepts the results of similar heart simulations as part of filings for drug licenses.

Researchers have also used computer simulations to reduce wait times at a hospital in Canada. Managers were able to model, in the virtual world, the effect of adding a walk-in clinic for people with minor injuries and adding more physicians at peak times. They then put what they had learned from the simulation into practice (Blake and Carter, 1996).

Nature can even be reengineered as biological systems are translated into information and information is translated into biology. As Richard Dawkins (2003) says, “Genetics today is pure information technology. This, precisely, is why an antifreeze gene can be copied from an arctic fish and pasted into a tomato.”

The ultimate form of experimentation is the emerging field of “synthetic biology,” radical techniques for making the genetic engineering of an organism work as routinely as the design of a new computer chip. One goal of synthetic biology is to design and build engineered biological systems that “manipulate information, construct materials, process chemicals, produce energy, provide food, and help maintain or enhance human health and our environment,” according to Drew Endy, a biologist at the Massachusetts Institute of Technology (Endy, 2005).

Unlike genetic engineering, which moves genes one at a time between species to create a new molecular entity, the goal of synthetic biology is to assemble genes from different organisms to create new metabolic pathways, and even new organisms. In the long run, this might involve rewriting genetic code altogether to build organisms that are beyond the capabilities of natural biology.

Experimentation in silico simulates existing systems. But synthetic biology is creating a new platform for investigating biological possibilities. Modeling the heart and modeling the hospital emergency room are akin to Sir Isaac Newton watching an apple fall and discovering gravity. Synthetic biology is analogous to Einstein’s theory of general relativity, which attempts to explain all of space and time.

The Adaptive Imperative for Health Care
An example of the adaptive imperative in health care is the Partners HealthCare physician order-entry (POE) system, which is already being used by the flagship hospitals in the Partners chain, Brigham and Women’s and Massachusetts General in Boston (Davenport and Glaser, 2002). The Partners system uses physicians’ experiences and outcomes to continually adapt and change the rules disseminated to physicians in the network.

To order a drug or a procedure or a test, a doctor must put the order into the POE system, which then looks at the patient’s information—the other drugs being taken and other important circumstances—and asks the doctor, “Did you know your patient is taking this drug, which is incompatible with what you just asked for?” Or it says, “Do you know that our board of cardiologists thinks that the new drug Avastatin is better than Lipitor?” The system doesn’t tell doctors what to do, but it gives them the latest knowledge, which can then change their decisions.

At the back end, the POE system collects information about patient outcomes, enabling the board of cardiologists to review them and revise their recommendations accordingly. Once they make a revision, every doctor in the network will see the revised rule the next time he or she logs in—just like the BMWs mentioned above.

When Partners experts determined that a new drug was helpful for heart problems and then made that information automatically available to doctors prescribing treatment, orders for that drug increased from 12 percent to 81 percent. As a result of the POE system, Partners has saved money because doctors are prescribing cheaper and more effective drugs and treatments, and hospital stays are shorter. In addition, adverse drug interactions have been reduced by 17 percent, saving about $10 million annually (Market Wire, 2002).

The health care delivery field has been an isolated ecology, and an ecology that is sheltered for too long becomes vulnerable to predators from the outside. New interloper species are already arriving, in the form of innovative companies like Steve Case’s Revolution Health and walk-in clinics at CVS and Wal-Mart stores, to destabilize the health care ecology and disrupt established business models We don’t yet know if those species will be invasive, like kudzu, or will be ill-suited to the environment and die out.

Navigating through the increasing turbulence will require adaptive leadership, rather than a command-and-control approach. Ultimately, an adaptive mindset will be necessary for health care executives to meet the increasing demands and for health care as a whole to compete with new species and shape the industry’s rapidly evolving transformation. The intensive use of simulation and other less hi-tech information tools—such as checklists—will be essential to ensuring that the evolution proceeds safely (Gawande, 2007).

As John Maynard Keynes wrote in The General Theory of Employment, Interest and Money (1936), “The difficulty lies, not in the new ideas, but in escaping from the old ones, which ramify, for those brought up as most of us have been, into every corner of our minds.”

Blake, J.T., and M.W. Carter. 1996. An analysis of emergency room wait time issues via computer simulation. INFOR 34(4): 263–279. Available online at ai_n8748764/pg_1.
Davenport, T.H., and J. Glaser. 2002. Just-in-time delivery comes to knowledge management. Harvard Business Review 80(7): 107–111.
Dawkins, R. 2003. A Devil’s Chaplain: Reflections on Hope, Lies, Science, and Love. Boston, Mass.: Houghton Mifflin.
The Economist. 2006. Life 2.0: synthetic biology. The Economist 380 (8493): 76.
Endy, D. 2005. Foundations for engineering biology. Nature 438(7067): 449–453. Available online at nature0 4342.html.
Gawande, A. 2007. The checklist. The New Yorker, December 10, 2007. Available online at fact_gawande/.
Gibbs, W.W. 2004. Synthetic life. Scientific American, April 26, 2004. Available online at 94F483414B7F0000&print=true.
Hall, S.S. 2007. TR 35: David Berry. Technology Review, September/October 2007. Available online at T&TRid=603.
Kauffman, S. 1993. The Origins of Order: Self-Organization and Selection in Evolution. New York: Oxford University Press.
Keynes, J.M. 1936. The General Theory of Employment, Interest and Money. London, U.K.: Macmillan Cambridge University Press.
Market Wire. 2002. Partners HealthCare Goes Live with Web-Based Physician Order Entry System Built on InterSystems’ CACHE. November 2002. Available online at mark02048422.
Meyer, C., and S. Davis. 2003. It’s Alive: The Coming Convergence of Information, Biology, and Business. New York: Crown Business.
Xie, F., Z. Qu, J. Yang, A. Baher, J.N. Weiss, and A. Garfinkel. 2004. A simulation study of the effects of cardiac anatomy in ventricular fibrillation. Journal of Clinical Investigations 113(5): 686–693. Available online at pubmedid=14991066.


About the Author:Christopher Meyer is chief executive of Monitor Networks. This article is based on his presentation at the NAE Annual Meeting Technical Symposium on October 1, 2007.