Biomedical Engineering Reference
In-Depth Information
Fig. 3 Components of an agent-based model. The grid of squares represents the simulation
space, which can hold concentrations of extracellular factors, denoted here by colored squares.
Agents/cells reside at discrete points on the grid, and are colored to represent cell phenotype or
activity. Agents perform behaviors according to their rule set. For example, the yellow cell is
currently static, with no outside influences, while the green cell is following a rule instructing it to
migrate up a gradient of a chemokine (purple). A third cell (red) is affecting the phenotype of its
neighbors through paracrine secretion of a growth factor (blue)
3.2.2 Agents
Agents are discrete entities that perform a defined set of behaviors according to a
set of rules. Agents representing cells, for example, are programmed to exhibit
biologically-relevant behaviors, including migration, proliferation, differentiation,
and apoptosis [
3
-
5
,
43
,
44
] as shown in Fig.
3
. Rules governing an agent's
decision to perform these behaviors may take into account, among other things, the
agent's past state history and factors present in the local simulated tissue envi-
ronment; these rules can thus be stochastic or deterministic. Agents can detect
biochemical and biomechanical stimuli in their environment and respond by
exhibiting a particular behavior. Agents can also affect their environment by
secreting diffusible growth factors (e.g., PDGF-BB) or by producing ECM and its
modifiers, such as collagen or MMP-9. Finally, agents can interact with one
another. For example, neighboring agents with engaged cadherins can interact
physically much in the same way as neighboring cells in a tissue would interact—
by signaling directly to one another, transmitting a mechanical force, and so forth.
All of these actions and interactions will impact an agent's state, and the ABM can
record the state histories of each agent at each time step, facilitating individual cell
lineage tracking. To simulate more complex phenomena, multiple cell types can be
represented using multiple types of agents. For example, an ABM modeling the
progression of an atherosclerotic plaque might contain agents that represent
individual ECs, SMCs, macrophages, foam cells, or fibroblasts. The aggregate of
agent behaviors and interactions over space and time produces emergent phe-
nomena that could not be predicted by modeling a single cell or using models
making a continuum assumption.
Search WWH ::
Custom Search