Building ARCHeS to Precise Answers, Stronger Decisions and Better Health

Oct 30, 2007, 1:05 AM, Posted by Risa Lavizzo-Mourey

Yesterday, I had the pleasure to publicly announce Pioneer’s largest grant to date: a $15.6 million award that will enable Dr. David Eddy and his team at Archimedes, Inc. to build ARCHeS, a powerful Web-based interface and delivery system that will allow far more health and policy decision-makers to use the Archimedes simulation model. That’s somewhat of a mouthful, so let me break it down a bit to describe why this project is exciting and has potential to transform the way that we make health and health care decisions.


So, what’s Archimedes? Perhaps the best place to start is with the problem it addresses. David Eddy—a rock-climbing heart surgeon turned mathematician and health economist—is a visionary driven to tackle the enormity of what we don’t know in medicine. So much of what guides our actions amounts to little more than enlightened guesswork. Simply put, there have always been far more questions than answers when it comes to providing quality health care.

As a doctor treating patients, a policy maker and, now, a foundation leader, I know this to be true.  Even the best-intentioned, most diligent among us in this field—be they doctors and nurses, health care analysts, researchers or policy officials—are continually frustrated by the gaps in information to guide many of our decisions. All too often, we can’t say with reasonable certainty what treatment...which policy...what mix of approaches... really works.

Mathematical Modeling: SimCity for Health Care

For more than 10 years, David, together with co-founder Dr. Len Schlessinger, has worked to develop Archimedes, a remarkably powerful, detailed and realistic mathematical model of human physiology, diseases and health care systems. It’s unique in many ways—it is the only model of its kind that can simulate pathways down to the cellular level in human physiology and, in turn, predict outcomes at the population level. By plugging questions into Archimedes, health care organizations and policy leaders can determine, reliably and quickly, the health and economic outcomes of specific interventions and treatments for patients, providers and payers.

As David states it, “We would never build a skyscraper, bridge or airplane without using models to optimize their designs and ensure their structural integrity. Yet today, that is what health care providers and administrators are forced to do with patient care, resource planning and other critical health care decisions.”

One way to visualize how it works is to think of SimCity for health care. What they’ve developed is essentially a virtual world of people—with different virtual physiologies, diseases, signs, symptoms, doctors, tests, treatments and outcomes—that health care and health policy leaders can use as a testing ground to answer questions that cannot be feasibly, safely or quickly studied in the real world.

There is a detailed, self-guided presentation on the Archimedes site that describes what the model does more fully.

The Method, and the Potential Impact

The model is fast, flexible and accurate. Using the same data as randomized clinical trials in many cases, it has reproduced or predicted outcomes that track nearly identically on to those same real-world trial results—in less time and at less expense. Archimedes is also the most detailed model of health care costs available, providing needed clarity on how different interventions may affect health care expenditures. There could be a big role for something like this in answering the key questions that dominate the health care debates of today—and tomorrow.

Admittedly, until you see examples of how this all works, it’s a bit abstract. One interesting case involves the American Cancer Society, American Heart Association and the American Diabetes Association. These groups jointly commissioned Archimedes to project the rise in cardiovascular disease, diabetes and costs that will be seen in the Medicare population over the next few decades, especially as the population ages and risk factors like obesity continue to worsen. The model calculated the effects on health and economic outcomes of 14 different evidence-based guidelines relating to smoking, cholesterol, glucose, diabetes exams, etc.

Archimedes’s results helped policy-makers determine the best ways to prevent increases in morbidity, mortality and costs associated with these specific diseases in this specific population. Specifically, the model forecast that full compliance with existing guidelines could reduce myocardial infarctions by two-thirds to three-fourths. Diving a bit deeper, however, they also learned that adherence to different guidelines triggers very different effects. Some should be given much higher priority than others.

And, here’s where it gets intriguing. David and colleagues estimate that if Medicare were to try to evaluate all of these guidelines using clinical trials—our traditional analytical gold standard—it would cost roughly $5 billion and take about 20 years.  Currently, using the Archimedes model, the analysis would cost several hundred thousand dollars, and take about six months to return reliable results that can inform real-world decisions.

The Disruptive Power of ARCHeS

When we first started talking with David Eddy about what would be needed to take Archimedes to the next level, he mentioned two key barriers: cost and complexity. Today, only highly trained scientists and mathematicians can use Archimedes. Each application of the model is specially configured, making it very costly to run.

What the Pioneer Portfolio’s grant will support is the creation of ARCHeS, a new Web-based interface and delivery platform for the Archimedes model that will be built and tested over the next five years. Ultimately, ARCHeS will permit far more health care decision-makers to access the model through the Internet.

For less money and with less training, I envision groups ranging from the Congressional Budget Office to the American Academy of Pediatrics to your state Medicaid office, using ARCHeS from their desks to plug in specific questions and apply the model’s results to guide their decisions. The cost and skill level required to use the model will come way down and, we hope, the number of users and breadth of questions answered will go way up.

Bringing this full circle, I want to revisit the example I gave above, in which the cancer, heart and diabetes associations applied the model to predict how 14 evidence-based guidelines fared in treating cardiovascular disease and diabetes. Archimedes returned clinical trial-like results in about six months, costing hundreds of thousands of dollars (better than billions, but still...). 

Using Archimedes via the planned ARCHeS interface, those groups could run the same analysis themselves in approximately three days and drive the cost down by perhaps 80 or 90 percent.

I’ve posted previously about how a truly disruptive innovation—as I believe ARCHes will be—makes it possible for people to have access to something they want, more easily, inexpensively and without having to rely on experts to the extent they did previously. In this case, ARCHeS will expand use of the model, lower the skill threshold to use it, and reduce the cost of use. End users—decision-makers in health care and policy organizations—will reap the most direct gains, though we hope that the advances in knowledge generated by their use of ARCHeS ripple throughout the health care system to eventually benefit individual providers, patients and consumers.

Why Pioneer?

I’ve long admired the work of David and his colleagues and the tenacity with which they have sought to strengthen and apply evidence-based medicine. I have little doubt that Archimedes – a for-profit entity – could gain venture capital support for a delivery system like ARCHeS. However, it is less likely that VC funders would place as much emphasis on ensuring that the model reaches the widest audience of public and nonprofit health care decision-makers. It also may be less likely that the cost of using the model would come down enough to put it within reach of groups with smaller budgets. This tool holds too much promise to remain accessible only to those with deep-enough pockets.

This, therefore, is to me a prime example of why a philanthropy like RWJF should support disruptive innovations like ARCHeS and the Archimedes model. Our grant will help the Archimedes team to build ARCHeS with the same skill and rigor applied to developing and improving the model itself. In 2011, when ARCHeS is slated to go live, we hope it becomes used widely by decision-makers to help them do their jobs more effectively. If this vision is realized, RWJF will have achieved an important goal – that is, to support powerful innovations that may transform the future health and health care landscape.

I’m proud that this grant will help realize our and Archimedes’ vision that the model be available to health care decision-makers throughout the nation and, indeed, the world. It will empower them to plan more effectively, target resources more precisely, lower costs where possible and, most important, deliver care that brings the highest benefit to patients. These are important and timely goals.

This commentary originally appeared on the RWJF Pioneering Ideas blog.