A health tool known as the Archimedes model has recently been shown to help predict the outcome of diabetes in patients. The Archimedes in the past has only been applied to large populations, not individuals, making this a novel study. It was also shown that the Archimedes model is equally efficient in predicting diabetes in patients as other tests designed specifically for this purpose.
The Archimedes model “is a large-scale simulation model of human physiology and health care systems. It has been extensively validated by its ability to quite closely replicate a wide variety of aggregate health outcomes in populations.” It is very intricate and complicated, and “uses scores of ordinary and differential equations to represent metabolic pathways, occurrence and progression of diseases, signs and symptoms, treatments, and outcomes.” Perhaps more understandably, Archimedes simulates the reality of a community or individual, incorporates appropriate medical data, runs, and generates a prediction based in its formulae that is thought to parallel what would happen in real life. This complex model has never been successfully validated for use on individual patients however, which is what the current researchers attempted to do regarding prediction of diabetes.
An extensive past study, known as the San Antonio Heart Study (SAHS) formed the basis for the current study. SAHS had been performed over an eight year period in which it followed 3,682 patients (62% Mexican, 38% Caucasian). SAHS also is used as a comprehensive predictive tool for individuals, and was compared against Archimedes by the current researchers. The current study selected 100 individuals at random from the study, 50 of whom developed diabetes at some point during the SAHS and 50 whom remained diabetes free.
All the necessary data was available from SAHS to implement the Archimedes model, which was done by plugging each individuals information into what the American Diabetes Association has called “Diabetes PHD,” a diabetes predictive tool run on the Archimedes model. The same data from the 100 individuals was also used to generate predictions using SAHS and another model known as Atherosclerosis Risk in Communities (ARIC) in order to compare against Archimedes predictive powers for diabetes risk.
SAHS and ARIC are individual specific tools desigened specifically to predict diabetic risk, as opposed to Archimedes, which has many functions. As the researchers explain it, “both the SAHS and ARIC models were built from person-specific data and optimized specifically for predicting incident diabetes. In contrast, Archimedes was designed to be used for a very wide range of purposes, calculates many different outcomes, was not built from person-specific data, and was not calibrated to determine the incidence of diabetes.”
The results of the study, even with the apparent disadvantage for Archimedes stated above, were very positive for Archimedes. The researchers concluded that the predictions of Archimedes strongly correlated with SAHS and ARIC, and that the differences in assessing the risk of developing diabetes between Archimedes, SAHS and ARIC “was not statistically different.”
Given that Archimedes serves a wide range of predictive functions, not just for diabetes, and not just for individuals, this study strongly supports the suggestion of implementing it as a common diagnostic or preventative tool. “This report extends the validation of Archimedes and demonstrates its excellent ability to discriminate between individuals who will or will not develop diabetes. Its utility is comparable with models developed solely for that purpose,” conclude the researchers. Further development of the Archimedes model in order to make it faster and more efficient is currently under way, and it is already available for use through the Internet for doctors and patients alike as a complimentary risk assessment tool.
Source: Defeat Diabetes Foundation: Stern, Michael. Williams, Ken. Eddy, David. Kahn, Richard. Diabetes Care. “Validation of Prediction of Diabetes by the Archimedes Model and Comparison With Other Predicting Models.” August 2008.