Living with Diabetes
Diabetes PhD and Archimedes
What it is Archimedes?
Archimedes is a math model that represents the anatomy, physiology and pathology related to diabetes and its complications. In the virtual world of Archimedes, every element corresponds to the same elements in the real world, one-to-one. There are virtual people, and each one has virtual organs and tissues (e.g., heart, pancreas, kidneys, eyes, blood). Like real people they have blood glucose levels, cholesterol, blood pressure, etc. Archimedes also includes all the other important aspects of a health care system, such as facilities (e.g., hospitals, emergency rooms, clinics, operating rooms), personnel (e.g. doctors, nurses) tests (e.g., lipid profile), and treatments (e.g., drugs, procedures).
Archimedes is like Second Life, where the virtual people all have virtual lives, buy virtual clothes and have virtual friends. In Archimedes each individual in the virtual world is simulated with a unique anatomy and physiology. Thus, the model is populated with thousands of simulated people, all of whom are living simulated lives, sometimes developing simulated diabetes and/or its complications. These people seek care at simulated clinics, are admitted to simulated hospitals, get simulated tests and treatments, and have simulated outcomes. This is all done at a high level of detail and accuracy.
The model is built using differential equations and object oriented programming. Descriptions of the structure and equations of Archimedes pertinent to diabetes and its complications have been published.(1,2,3) The use of differential equations preserves the continuous nature of biological variables as well as the interactions between them. The model is continuous in time: biological variables are changing and interacting continuously, the natural histories and severity of conditions progress smoothly, any clinical event can occur at any time, and the timing of events are as condensed or drawn out as occurs in reality. Interventions, both to prevent diabetes and to manage it when it occurs are modeled at the level of the underlying biology.
The model includes virtually all of the biological factors related to diabetes and its complications, including A1C, insulin resistance, beta-cell function, atherosclerosis, eye, kidney, and heart functions and their complications, and the effects of all relevant drugs and other key therapies (e.g., weight loss) on organ and tissue function.
How do we check its accuracy?
The ability to create simulated people who represent real people provides us with an opportunity to confirm the accuracy of the Archimedes model. Specifically, we can take the people who are entered into real clinical studies, create simulated versions of those people where the simulated population matches the real population with respect to all the important variables (the types of variables users of Diabetes PHD or My Health Advisor are asked to provide), and then use Diabetes PHD or My Health Advisor to calculate the outcomes that occur to those simulated people over time. This has been done for about two dozen of the most important clinical studies that pertain to diabetes and its complications. Furthermore, within each trial there are often several different outcomes that enable us to test different part of the model (e.g., the physiology of the heart, kidneys, eyes) giving a total of almost 100 separate tests of the model's accuracy. The Archimedes model has proven itself to be extremely accurate, with nearly a perfect correlation between what the model predicted for the simulated people compared to what happened to the real people in the studies. In most of the validations, no information from the real study was used to build Archimedes. Indeed, Archimedes was able to predict correctly the results of studies that had not even been concluded at the time of the prediction.
How does Archimedes work?
When a person uses Diabetes PHD or My Health Advisor and provides information about himself or herself, Archimedes creates a simulated person who has the same characteristics (e.g., sex, age, race/ethnicity), same features (e.g., height, weight, blood pressure), same laboratory test results (e.g., glucose, cholesterol), the same past medical history, family history, symptoms, complications, and the same treatments as the person providing the information. Archimedes then takes this simulated version of the person and creates a thousand "identical looking" people. It then calculates a future for those people–what is going to happen to them over the next 30 years. Because there are a lot of chance events that affect our futures, it is not possible to predict exactly what will happen to any particular individual. For example, it is not possible to predict that a person will have a heart attack precisely on May 17, 2012. However, by calculating a future for many people like the person who is using Diabetes PHD or My Health Advisor, we can "average out" those chance events and calculate the likelihood that important events like heart attacks will occur at various times in the future. Thus, while we cannot say that a person will have a heart attack on any particular day, we can say for example, that there is an 80 percent chance the person will have a heart attack sometime in the next 5 years. These probabilities are displayed as graphs and reported back to the user as their personalized results. Archimedes also calculates what would happen, if on the very day the user describes their current condition, they decide to improve their features (e.g., lose weight, quit smoking) from that day forward.
- More information about the Archimedes model.
- See a list of research studies that Archimedes incorporates to help power Diabetes PHD and My Health Advisor.
_________________________
1 - Schlessinger L, Eddy DM. Archimedes: a new model for simulating health care systems - the mathematical formulation. Journal of Biomedical Informatics 35: 37-50, 2002.
2 - Eddy DM, Schlessinger L. Validation of the Archimedes Diabetes Model Diabetes Care 26: 3102-3110, 2003.
3 - Eddy DM, Schlessinger L. Archimedes: A trial-validated model of diabetes Diabetes Care 26: 3093-3101, 2003.
