We are hosting a visitor this week in Morgantown: Alan Mills, a physician, actuary, engineer, and complex systems design person who applies very interesting tools to solve the problems of health prediction and healthcare design.

His toolbox includes using a technique called agent-based simulation modeling that can assess which elements of a system are most important for outcomes of the group. By giving each element in the model a unique identity and set of rules to follow (agents can be people, stores, hospitals, clinics, houses, etc.), the model can run simulations and give outcomes for these agents.

By changing the rules of the agents, you can test in a computer environment what might be the outcomes of changing this single variable in a real life setting. This can help you understand which elements have the biggest impact on the outcome of the network.

This is called perturbance of the system and is the only way to determine the elements most important for the outcome of a complex network.

When you change the elements or the rules, there are two outcomes: change in the outcome of the system or no effect. If a small change in the rules for one element has a measurable impact on the function of the entire network, that indicates the element has a very important function in the complex network.

Why is this approach needed?

Experience tells us that simple solutions to complex problems are rare. Incentives that reward certain individual behaviors fail to recognize the interconnectivity of the group of elements in a system, leading to the classic unexpected consequences of a decision.

I wrote about the law of the few, and in complex systems, each element in the system has a different contribution to the function of the whole. The most connected parts of the network are called hubs, and there are few of these.

Understanding this phenomenon has practical value. Let’s look at the impact of social networks.

In a groundbreaking study, Nick Christakis and colleagues found that individuals are 57 percent more likely to be obese if they are in a friend group with someone who is obese. Importantly, if the obese person was identified by the thin subject as a friend and the feeling was reciprocated, the risk of obesity increased by over 100 percent for the thin subject.


This approach can be used to determine factors most important in individual health, healthcare delivery, or healthy community creation.

We are using these tools to understand which elements in our state of West Virginia are most important to helping realize our why:

Tangibly improving the lives of West Virginia's citizens, elevating their lives and the state by creating one WVU and one West Virginia.