Sepsis is complex because outcomes are not only life and death. While survival occurs in approximately two-thirds of patients, other patients can die with a course of disease that has different hallmarks. Our simulation attempts to capture these outcomes.

Survival: The pathogen is destroyed and the patient’s damage levels return to near zero. This outcome usually results from a low rate of pathogen growth and a higher collagen production rate.

Persistent Infection: Pathogen levels and damage levels continue to rise unchecked by any other process. This occurs with a high rate of pathogen growth and a high rate of collagen production.

Aseptic Death: Pathogen has been cleared away, but damage continues to persist. This is a non-abating inflammatory response. Usually, aseptic death comes from low pathogen and collagen growth rates.

Septic Death:Pathogens and damage have gone out of control. Infection and inflammation continue to proliferate unchecked. The model shows this result when both the pathogen growth rate and collagen production are high.

Adjusting the different slides produces different patient outcomes. Try out the set parameters that result in survival, persistent inflection, aseptic death, and septic and then experiment your own parameters. You are the doctor, and your challenge is to treat your patient (that is, to minimize the “damage”).


This is a model of the inflammatory response. It implements pathogens, macrophages, and fibroblasts, and the chemicals LPS, collagen, and pro- and anti-inflmmatory cytokines.


The different types of cells are agents, and the chemicals are patch variables. The patches themselves are the tissue. Inter and Intra-cellular interactions are defined by the Agent-based model rules that resemble their biological nature. Different levels are shown in monitors and graphs.


After choosing beginning conditions and constants using the sliders, one should press Setup, and then press Go. Go is a continuously running function, and should be pressed again when one wants the model to stop. To reset the model, one needs to press Setup again.

The different setting buttons (in CAPITALS) preset the sliders to give a situation where that type of consequence would occur.


The user should notice how the different parts of the model interact, and the various graphs.


The user should first try to play around with the first group of sliders (on the top). These change the inital number of different cells. After tries of that, the user can go on to play around with other settings. The sliders are ordered in groups from top to bottom, in the order of the constants that have the largest effect on the model.


The model can be extended to have more exact interactions with the introduction of other chemicals.


The agent and patch functions of NetLogo really help facilitate an Agent-Based Model. The interactions and intuitively defined in the procedures.


Gary An's ABMs simulate similar topics wondefully.


Done by: Yakov Pechersky, Keely Rasmussen, Qi Mi, and Yoram Vodovotz.

Thanks to: Jack Kent Cooke Foundation, University of Pittsburg, Society of Complexity in Acute Illness, Gary An, and many others.

view/download model file: AMP2008_3.nlogo