MIDAS Steering Committee Meeting Minutes

May 6, 2008
Emory Conference Center
Atlanta, GA

Next Meeting: May 4, 2009
Interim Meeting of Outreach Subgroup: June/July, 2008 (date to be determined)


Steering Committee Members

Don Burke
Robin Bush
Blake Caldwell
Irene Eckstrand
Stephen Eubank
Harold Garner
Kathleen Gensheimer
Eric Jakobsson
Roz Lasker
Ira Longini
Ellis Mckenzie
Michael Merson
Alan Perelson
Bob Pinner
Richard Platt
Marc Lipsitch
Gary Smith
Diane Wagener

Other Attendees

Phil Cooley
Josh Epstein
Steven Frank
Susan Huang
Bruce Lee
Joseph Pratt
Doug Roberts
Jen Villani
Susann Westlake
Bill Wheaton

Goals for 2008

  • Continue to develop collaborations, such as specific groups within the Centers for Disease Control and Prevention, the Society of Healthcare Epidemiologists, the Agency for Healthcare Research and Quality, the National Institute for Allergy and Infectious Diseases, and others.
  • Continue to develop and implement MRSA projects.
  • Continue studies to compare and characterize model structure and behavior. 
  • Continue to provide materials and data that would be useful to the entire program.
  • Develop tools cautiously, in collaboration with users, and with attention to good design and utility.
  • Study the policy implications of models.
  • Extend models to address institutions (e.g., hospitals, businesses), clinical behavior, and individual / social behavior.
  • Consider studies of tradeoffs and comparisons of various policies and responses.
  • Consider developing tools that individuals can use to assess their own risks.

Review of 2007 Priorities

1. Develop hierarchies of models from analytic to highly-detailed agent based models.

MIDAS has developed and made use of a variety of modeling approaches, including, for example,

  1. Equation-based studies of antibiotic resistance (Lipsitch)
  2. Docking of models to data sets (Burke)
  3. Use case model of infectious disease dynamics (Wagener/Cooley)
  4. Construction of simple models to interrogate large agent-based spatial simulations (Smith/Tildesley)
  5. Large, spatially explicit agent-based social network models, including algorithms for computing rapidly on those networks (Eubank)

2. Expand modeling to other infectious diseases, including enteric and vector-borne.

MIDAS investigators are examining a range of diseases, for example,

  1. Docking model of enteric disease (cholera) to real data, including transmission from a common source and person to person (Longini)
  2. Vaccine trial designs for dengue (Longini)
  3. Vector model for dengue (Longini)
  4. Salmonella Newport spread in agricultural animals (Smith)
  5. Incorporation of spatial models in studies of malaria (Eubank)

3.   Develop research in incorporating social behavior, evolution, and ecology.

MIDAS has held four workshops (Social Networks, Social Behavior, Trust and Compliance, and Evolution) to gain a better understanding of how to incorporate these elements into models.  Progress in incorporating these areas into modeling is proving difficult. 

4. Data acquisition and management is a high priority for the network.

Focus on identifying data sources from past outbreaks, including international data, and on making these data readily available to researchers.  Researchers must be actively involved in this process.

The Historical Data and Document Catalog (HDDC) contains historical data on 1918, 1957, and 1968 influenza pandemics; a collection of CDC and WHO epidemiological reports; 150 geospatial data sets; and some international data.  Synthetic population data are available on request. 

Several MIDAS PI's noted that they need to submit material to the HDDC.

Discussion focused on the difficulty of getting good data on outbreaks in human and agricultural animal populations.  For example, HIPAA requirements may apply for geospatially explicit data.  Collaborations with state and territorial epidemiologists and with associations of health care providers may be useful.

In response to a question about usage, Dr. Wagener noted that the resources, which were requested by MIDAS groups, are underutilized. 

The steering committee reiterated that the research groups need to both submit and use available data from the HDDC.

5.   Continue and expand research on emergence and spread of antimicrobial resistance.

MIDAS research groups have expanded into these areas.  For example,

  1. A coalition of MIDAS investigators, CDC, and AHRQ is collecting data and modeling various aspects of MRSA.
  2. Marc Lipsitch, in collaboration with Gabriel Leung in Hong Kong, is studying antiviral resistance in Pneumococcus strains.
  3. Ira Longini, in collaboration with Rustom Antia, is studying the dynamics of the evolution of resistance.

6.   Continue to develop relationships with relevant agencies and organizations.

Blake Caldwell noted that MIDAS has passed the tipping point in collaborating with CDC.  Several CDC research communities are interested in collaborating with MIDAS investigators.  In addition, NIGMS and AHRQ have agreed to collaborate in a MRSA study, with AHRQ collecting data and MIDAS developing models of MRSA flow in hospitals and long-term care facilities.

7. Deposit all data, results, and models to the MIDAS model repository in compliance with the MIDAS policy on intellectual property and data sharing.

Phil Cooley from the Informatics Group pointed out that the model repository contains 14 models from research groups, including source code for 8 of those.  The Global Epidemic Model and TranStat are available on the Portal.

Site Visit Review

The steering committee reviewed the summary site visit report and briefly discussed the visits to RTI, Harvard Medical School, Harvard School of Public Health, University of California at Irvine, and the University of Pennsylvania.  The steering committee was enthusiastic about the cross-project and multi-faceted MRSA studies and for the research on simple models to supplement large simulation models.  In particular, the steering committee encouraged research groups to work with the informatics group to develop useful, robust, and available software through a software engineering process.

DHHS Home Antiviral Stockpiling Project

MIDAS has been asked to do some modeling with respect to a specific plan to encourage the purchase and stockpiling of antiviral medications in homes in advance of a pandemic.  There was considerable discussion about implications of the plan proposed by DHHS.  The steering committee recognized the importance of MIDAS's participation since it could raise critical points and add valuable knowledge.  For example, MIDAS could address whether 25 percent coverage, as proposed by WHO, is adequate and study the implications of inequitable distribution of antiviral medkits.  The steering committee suggested that MIDAS consider providing information on effective and efficacious policies, in addition to examining the implications of the particular policy under discussion.

MRSA Project

Susan Huang, Bruce Lee, and Gary Smith presented various aspects of MIDAS's new group project on the dynamics of MRSA.  A substantial part of the project is being conducted in collaboration with CDC, AHRQ, and health care organizations in Orange County, CA.  Models being developed for dynamics in agricultural facilities may prove relevant and useful to studying patient and MRSA flow in human institutions such as hospitals and long-term care facilities.

Presentations from the speakers are available on the MIDAS Portal.

Model Comparison and Simplification

Stephen Eubank, Marc Lipsitch, and Diane Wagener presented MIDAS efforts to compare the performance of models and to create simple, fast models of complex dynamics. Independent development of models, followed by comparison and characterization, is optimal.  The goal is not to make all models come to the same conclusion, but rather to discern why models differ.  This is an exercise in hypothesis testing. 

The study on targeted layered containment carried out in collaboration with the Homeland Security Council was MIDAS's first explicit effort to compare models and pointed out differences in model structure, interpretation of parameters and interventions, and calibration. 

The steering committee encouraged MIDAS to devise ways to characterize and compare models, in large part because when there is a need, we can apply the appropriate modeling tools.  A hierarchy of models is an important goal, in part because complex models need simple counterparts and explanations.  Likewise, the process for freezing and storing models periodically is critical.

Tool Development

MIDAS has undertaken development of tools for particular purposes.  TranStat is being developed to detect human to human transmission of disease and to collect critical data early in an outbreak.  Gary Smith's group is developing a tool for running tabletop exercises that takes account of the stochastic nature of outbreaks and a variety of specific decisions. 

The steering committee advised MIDAS to reach out to user communities and build tools in response to needs.  MIDAS should continue to study how to disseminate tools (e.g., Web-based, download).  It is also critical to make clear the uses of software tools and to have in place a system to identify problems and programming errors.


The new MIDAS booth for professional meetings and new outreach materials were presented at the meeting.  Outreach materials describe resources (e.g., synthetic data, HDDC, model repository) that are publicly available.  The informatics group has also undertaken an evaluation of the MIDAS Portal, the most obvious way that MIDAS presents its identity and mission to the scientific and policy communities and to the public.

The steering committee urged caution in moving too fast or too far with outreach activities.  Because of limited time for thorough discussion, the recommendation was to set up a separate meeting of steering committee members to discuss outreach activities.  Members expressing interest are Blake Caldwell, Skip Garner, Eric Jakobsson, and Kathleen Gensheimer.

The meeting adjourned at 4:00 pm.