Title: Professional profiling of physicians with writing behavior observed in electronic health records

Co-authors: Anastasia A. Funkner, Sergey A. Fokin

Abstract:

One of the most important and hardest to model factors in human lives is the influence of their social network. People interact with and influence each other, but they also change over time and so does their social network. Understanding the interactions and evolution of their social structure is complex but vital for research in pressing topics of human behavior such as depression and substance abuse. Other social and societal topics such as opinion dynamics and segregation too are very much impacted by the dynamic social environments. We are developing a python package to ease complexity scientists to model dynamic interacting social networks with complex agents. Every heterogeneous node/agent can be a model on its own, with its own variable continuous parameters, or states. They can influence and have flexible interactions with their peers. Furthermore, the agents are able to change their social network by changing their connections in strength and breaking or creating new links, making the framework able to model network dynamics. Using standard packages such as networkX and built from the ground up to provide flexibility, speed and analytics, it eases the network modeling as much as possible to help the user focus on the actual internal models and defining the interactions. It is possible to take existing models where outside factors are variates and investigate results on a network level. In our talk we will discuss the possibilities of the framework and give examples, aiming to inspire complexity scientists to apply this dynamic network agent-based method on their own field of expertise.