Could Electronic Trials Provide Better Evidence?
A mathematical model can expand on clinical results quickly and cheaply. But physicians may be skeptical.
Just as he evangelized about the need for physicians to base medical decisions on empirical evidence, physician and mathematician David Eddy is now promoting a mathematical model to fill the evidence gaps of the movement he helped start 30 years ago. The model, which he calls Archimedes, is built from hundreds of equations that represent human physiology and the effects of disease, and is overlaid with hundreds more simulating tests, treatments and physician behaviors. Archimedes expands upon clinical trial data to answer questions that trials can’t, Eddy says.
For instance, it can project how changing the dose of a drug or combining it with other medications may affect patient health. It can also estimate how a broader population than the one studied might benefit from a new intervention.
What’s more, Archimedes can be much faster and cheaper than a clinical trial. In two months, at a cost of several hundred thousand dollars, Eddy was able to replicate a seven-year, multimillion-dollar controlled clinical trial studying whether the drug atorvastatin could prevent heart attacks and strokes in diabetics—and Archimedes was almost spot-on in predicting the trial results before they were announced.
The model can’t replace clinical trials, however. “First you need empirical evidence from a trial that a test or treatment is effective,” Eddy says. Only then, after Eddy has developed an accurate model based on those results, can Archimedes apply it to untested applications.
But if physicians find fault with guidelines created from actual trials, why would they trust a model? “All we can do is show that the model can reproduce what is seen in reality,” says Eddy.