Skip Over Navigation Links


Because of a lapse in government funding, the information on this website may not be up to date, transactions submitted via the we​bsite may not be processed, and the agency may not be able to respond to inquiries until appropriations are enacted. The NIH Clinical Center (the research hospital of NIH) is open. For more details about its operating status, please visit Updates regarding government operating status and resumption of normal operations can be found at

MIDAS Consultation on Modeling Evolutionary Processes

October 22-23, 2007
University of Pittsburgh
Pittsburgh, PA


James Anderson
Georgiy Bobashev
Don Burke
Phil Cooley
Marcelo Costa
Derek Cummings
Jay Depasse
Irene Eckstrand
Steven Frank
Patrick Grim
Betz Halloran
Gabriel Leung
Lingling Li
Ira Longini
Les Real
Steven Riley
Roni Rosenfeld
Gary Smith
Jonathan Sugimoto
Yang Yang


The NIH / MIDAS network has developed computational models of epidemic infectious diseases that have been central to pandemic planning and policy development by the US government, the WHO, and other agencies world-wide. However, one limitation of the current generation of MIDAS epidemic models is that they do not include any representations of microbial evolution. The main objective of this workshop was to consider how best to conceptualize, represent, and code viral evolution into large-scale epidemic models. We will focused especially on influenza, but also considered epidemic and emerging disease threats more generally.


Consultants to MIDAS introduced aspects of viral evolution that could be considered in models.  The presentations are available to the MIDAS network on the MIDAS portal.

Characterizing intra-host diversity of influenza A virus and its effects on evolutionary dynamics
Elodie Ghedin (University of Pittsburgh)

The evolutionary genomics of influenza A virus
Eddie Holmes (Pennsylvania State University)

Infective dose, antigenic variation, and natural vaccination
Steven Frank (University of California, Irvine)

Spatial dynamics and molecular evolution of rabies and ebola
Les Real (Emory University)

Network perspectives on microbial ecology and evolution
Lauren Meyers (University of Texas at Austin)

Dynamics and in vivo evolution of viral infections
Dominik Wodarz (University of California, Irvine)


The group identified a number of open issues related to including and representing representation evolutionary processes in models of the transmission of infectious diseases.    The overarching questions were “Is it possible to include evolutionary processes in large-scale models?  If so, how? Recognizing that the time scales of evolution and policy making are quite different, there is a limited number of questions that can be addressed.

The group tried to identify problems that might occur on a fast enough time scale for evolutionary processes to have impacts.  These included antimicrobial resistance, emergence, and endemic diseases.  The first two of these were discussed further.

Some participants suggested that the process of emergence is inherently stochastic and difficult to obtain data on, and so efforts might be better spent in other areas.  Some of the MIDAS effort addresses questions at a time-scale where we might not expect evolutionary pressures to act, for example, the models of the initial spread of pandemic influenza.  The question was posed to those in the room familiar with influenza - Are our predictions about the characteristics of a newly emergent flu virus reasonable and can we assume that evolutionary processes will not be important once the jump to humans has been made?  Some suggested that existing approaches (see work by Antia, Bush, and Ferguson) have successful in describing the processes of emergence of a new agent.

Rather than modifying transmission models to include the impact of evolution, a suggestion was made to use some of the approaches taken in transmission modeling (e.g. individual based modeling) to study evolutionary processes.

Drug Resistance and Vaccine Escape
Drug resistance is an area with many open questions and important in almost any pathogen system in which drugs are used. Particularly in influenza, there is interest in evaluating how rapidly viruses will escape a given vaccine.  It may be particularly interesting to quantify the economic consequences of particular policies incorporating costs associated with the loss of effectiveness of a particular intervention due to resistance or escape and the uncertainty associated with this outcome.

The subject raised the issue of whether modeling antigenic evolution would be useful in designing vaccines.  Influenza is perhaps the best system in which to study evolutionary processes because of the amount of data, and understanding of mechanisms.  Thus, flu and the question of vaccine design might be one of the best areas to focus MIDAS’s efforts.  On the other hand, genetic change does not imply antigenic change.  The mapping of genetics to antigenicity is not well understood in most systems.  We cannot yet predict genetic or antigenic changes in influenza viruses from year to year or estimate the frequencies of various strains.

Relevance to Policy

Policy tends to deal with causality, but evolutionary processes rarely have that kind of certainty.  It is difficult to translate detailed modeling results into simple explanations of phenomena for policy makers.  A familiar example of doing this well was the climatologists telling Bill Clinton that he would not have maple syrup on his pancakes if global warming continues. 

One approach would be to identify decisions that will have to be made by policy makers in the near future and develop models to address them. Recommendations meant to inform policy should consider strategies that lay out rules of designing future strategies in order to quickly respond to information about an emergent issue.


Evolution has a strong theoretical component, but the tradition of theoretical approaches in epidemiology and biology is not as strong as in other fields such as physics. Federal agencies and individual institutions should promote the use of theory. The development of useful heuristics was also viewed as a key way to increase the acceptance of theoretical results.

The complexity of modeling results must be reduced in order to facilitate interaction between field and empirical workers and theoreticians.  A close relationship between theoreticians and empiricists is very important for the further development and success of theoretical approaches. 

What features of evolution should we capture?

The group identified the following as useful:

  • Dynamics of co-infection
  • Spatial organization of information (sequence, disease, …)
  • Coexistence of variants in ppn long-term
  • Strain interactions estimation of parameters
  • Descriptions of systems (transmission, etc.)
  • Impact of immune response on treatment outcomes

What are important evolutionary questions?

The group identified a number of problems or systems fruitful places for MIDAS to focus on. 

Methicillin-resistant Staphylococcus aureus is an important problem in the US and multiple open questions could be studied.  For example, the importance of community versus hospital settings for transmission, the comparative phylodynamics of community and hospital-acquired MRSA, and the impact of treatement programs organized at multiple spatial scales (hospitals vs. cities vs. national) would be interesting.

Vaccine coverage/polio
Theoretical work could help direct polio eradication efforts.  There is a tradeoff between using the more effective oral vaccine and the increased probability of emergence of vaccine derived transmissible polio viruses.

Evolutionary social behavior
The interaction between behavioral processes and evolution of pathogens is an interesting area of research.

Network structure influence on evolution
Host network structure may have important impacts on the evolution of pathogens.

Evolution of different systems such as linguistics, genetics, disease, computation
The representation of other systems undergoing evolution might provide useful insights into how to represent the evolution of pathogens.

Explicitly multiscale models

  • perturbation
  • time/policy scale
  • careful not to be a model of everything


Antigenic variation-temporal and spatial and species

Flu drug resistance


Evolution of pathogen in multiple hosts (malaria)

Models-data analysis around hospital acquired resistance and competition between strains

Models as tools for statistical analysis

This page last reviewed on November 14, 2014