Health care decision-makers have precious few cost-effective sources to look to for help in answering health care questions. For example, a state Medicaid director wants to determine which illness prevention activities are the most effective over a 30-year period. This question is far too complex to address using conventional methods. Yet an accurate answer would enable Medicaid and other agencies to better serve those enrolled in their program—and generate massive savings in time and money.
Archimedes is a remarkably accurate software model that incorporates proven mathematical representations of physiological processes relevant to diseases into a framework of clinical, administrative and other "real world" influences. At a very basic level, the Archimedes model creates virtual people who have virtual physiologies, experience virtual signs and symptoms, go to virtual doctors, get virtual tests and treatments, and have virtual outcomes. Health care decision-makers can use this simulated world as a testing ground to answer health care questions that cannot be feasibly, safely and quickly studied in the real world.
Its accuracy has been thoroughly validated—Archimedes staff have taken the data used in more than three dozen clinical trials, simulated those trials in the model's "virtual world," and returned results mirroring those of the real world trials.
"The model has proven to be far more accurate than anyone, including us, ever expected," says David Eddy, M.D. Ph.D., medical director and founder of Archimedes, Inc., the company responsible for the Archimedes model.
The problem is accessibility. At present, only highly trained experts know the model's code, meaning that each application of the model is hand-configured and costly to run. "It has a good interface that we use," says Eddy, "but we need to create one to enable people outside the company to use it, simply and without errors."
RWJF is funding a $15.6-million, five-year development effort to make the Archimedes model more accessible to health care decision-makers. ARCHeS (ARChimedes Health care Simulator) will make the model "very accessible," says Eddy. "It will have a Web portal," making it internet-ready. "It will be very intuitive. We'll have a training program, manuals and [telephone] help lines." Eddy is envisioning that "organizations that make decisions have someone on staff" who will be trained to use ARCHeS. He anticipates that "anyone who has a master's degree in physics or statistics should be able to use the program easily."
Fully implemented, ARCHeS will dramatically increase the pool of potential users, meaning decision-makers will be defining target populations, possible interventions and outcomes online. They will simply click on "calculate" and get the results in minutes, at far less cost. In addition, they can interactively vary inputs to see how different scenarios play out, secure in the knowledge that model results are based on the best available research data.
With this increase in users, however, comes the risk of misuse, as well as privacy concerns. To reduce the possibility of user-induced errors, the project will "create templates with information on [commonly used] medical populations, such as the Medicare population," says Eddy. "That way users won't have to enter as much data." They will also incorporate such data sets as national guidelines and practice patterns for commonly encountered problems, and allow users to create their own templates. "We will invite people to share templates" via a library, says Eddy, with the templates vetted by ARCHeS staff for accuracy.
As for privacy protections "there will not be patient-specific data anywhere in the system," according to Eddy, and "user's information will be protected by the highest standards of privacy."
Eddy also expects to encounter some skepticism as ARCHeS gains more users, but believes the nature of the Archimedes model will allay concern. "We are starting deep in the biology and physics," he says. "We are building a natural model at a level of detail physiologists and physicians understand."
Broadening ARCHeS' accessibility is a major goal, but it must be market-worthy to become fully integrated into health care decision-making. Market discipline is ensured by a simple mechanism—Archimedes is a for-profit company, although it is wholly owned by the nonprofit Kaiser Permanente Southern California. "It will be tested in the marketplace, which is one of the best ways to ensure that it provides value," says Eddy. Any profits "will be plowed back into the model. We have to keep it up to date—and the science is changing fast."
"Archimedes is a very powerful simulation model but it is very technical and expensive to use," says Nancy Barrand, M.P.A., RWJF special adviser for program development. The ARCHeS proposed "software interface will make it far more available and affordable. We want to apply the power of mathematical modeling to help usher in a new era of health care decision-making."