Patient Registries, Predictive Models, and Optimal Care

To the Editor: In their Commentary, Drs Dreyer and Garner1 extolled the virtues of real-time data from patient registries as an important supplement to data derived from clinical trials. They advocated for methodological research to increase understanding of what constitutes quality and a more directed effort to evaluate the strengths and limitations of different types of evidence. However, they did not acknowledge powerful predictive modeling tools that are available to transform real-time data into evidence defining optimal care. The Archimedes model, for example, is a versatile and powerful mathematical model enabling prediction of clinical outcomes based on signs and symptoms of disease, patient demographics, laboratory data, and even physiologic variables. It can provide cost estimates for various care management options to optimize cost-effectiveness ratios. A work in progress, the model has been rigorously validated for diabetes.2,3 Such a virtual health care system could be used not ..