Needless to say, sepsis pathophysiology extends beyond the isolated and static models mentioned above, and involves changes over time and between tissues. Several methods can be used to capture and analyse such spatio-temporal dynamics [
15]. Pigozzo and colleagues specifically modelled the response of the innate immune system to lipopolysaccharide (LPS, a bacterial product) in a dynamic computational model based on partial differential equations [
16]. By including several cell types, pro-and anti-inflammatory cytokines, and the diffusion between the vascular system and tissues, the model is able to reproduce features like the temporal influx of specific cell types and the cytokine-mediated resolving of inflammation. More recent work expanded on this concept, with the incorporation of clinical, patient-derived data in order to validate computational results [
17]. A different approach to incorporate spatial and temporal dynamics is the use of cellular automata, which are abstract collections of cells that have distinct states and are organized within a grid of finite number of dimensions. The states of these cells can be updated over time, in a discrete or probabilistic manner (called stochastic cellular automata), and thereby recreate interactions and organizing behaviour that can model the immune system [
18,
19]. Such approaches have been used to model the microenvironment of the lung during tuberculosis infection [
20], study the balance between necrosis and apoptosis in neutrophils during inflammation [
21], or analyse the concerted behaviour of cell populations [
22]. It is also possible to model the interaction between different sub-systems. In a two-layer model of inflammation—in which the innate immune system and parenchymal cells each oscillate, and indirectly interact through cytokines—it was shown how the complex dynamics between layers could lead to a healthy synchronized state, or a pathological state of the parenchyma [
23]. Such analyses provide information on how the dynamics between sub-systems can determine state-transitions, which in a clinical sense could for instance translate as organ failure in a patient with sepsis. Network-based analyses, in which features are often represented as nodes, and connections as edges, also play a key role in understanding and visualizing complex systems. Networks can be constructed at multiple levels: for dynamic molecular intracellular pathways [
24], for crosstalk [
25] and interactions [
26] between cells or populations of cells, or to represent the spatial organization within tissues [
27]. Mostly, the combination of -omics technologies and network-based analyses has enriched our understanding of the molecular basis of immune activation on an intra- and intercellular level [
28,
29]. Herein the surge in availability of high-dimensional datasets may allow for the use of machine learning techniques [
30], although the “black box” nature of many of those methods makes it challenging to actually derive at functional and mechanistic insights. A combination of network-analyses with computational models, thereby mathematically defining the edges between the nodes and quantifying these interactions, could form the basis for predicting how the host might respond to infection. For instance, a recent “whole-body” computational model reproduced how systemic inflammation can cause relative hypovolemia through endothelial hyperpermeability, and the effects of fluid administration and norepinephrine hereon [
31]. The model also showed differential patient outcomes when certain interventions were introduced, illustrating the potential translational and predictive value of such models. Together, these methodologies and studies each reflect considerable progress in analysing complex adaptive systems. Ideally these methods should complement and enhance each other, rather than be used in isolation [
32]. One example is the combination of artificial neural networks and agent-based modelling, which has been used to predict cytokine trajectories and disease progression in (virtual) patients with sepsis [
33].