Proposal to Reannounce the NIGMS MIDAS Program

NIGMS Advisory Council Meeting
January 24-25, 2008

This document summarizes staff and outside reviewers' evaluations of the current MIDAS program, presents the case to continue the program, and recommends specific changes. Staff requests permission to reannounce the modified program in the form of three Funding Opportunity Announcements.

BACKGROUND

The NIGMS Models of Infectious Disease Agent Study (MIDAS) program was launched as an RFA in November 2004 as a response to a national need. The 9-11 tragedy, subsequent postal anthrax attack, and concerns about use of smallpox as a bioterrorist agent highlighted the vulnerability of the nation to terrorist attack, including the use of infectious biowarfare agents. The Fogarty International Center sponsored workshops on the role of epidemiological modeling in addressing infectious disease (ID) threats, including threats of naturally emerging infectious diseases. Simultaneously, NIGMS's new Center for Bioinformatics and Computational Biology was formulating its mission, which included developing models of biological processes and dynamics. The director of NIGMS initiated a program that would leverage NIGMS's interests in computational biology and the evolution of infectious diseases, to develop a consortium of laboratories engaged in modeling infectious disease dynamics. The goals of the program were two-fold. The first goal was to support cutting edge research in three specific areas:

• Modeling of Host-Pathogen Systems
• Modeling of Disease Systems
• Modeling of Response Systems

The second goal was to provide for coordinated, expert consultation to policy makers in the event of an anticipated or actual national emergency in which modeling of outbreaks would be important. The inclusion of this "operations" component was novel for NIGMS, and was adopted from the perception that such organized consultation was currently lacking and would be essential in the face of threats.

The intent was to fund pilot projects in the above research areas that would rely on analyzing existing data; generating new infectious disease or epidemiological data was not included in the original scope. In addition to the research groups, the program called for a separate informatics group to provide computational capacity and informatics support to research groups. Staff envisioned the informatics group as responsible for developing and managing databases; analyzing and verifying models; identifying and documenting relevant data; and creating a portal for access to models, tools, and other information.

The MIDAS consortium began with three research groups and the informatics group in May 2004; the RFA was issued again in 2005, and four additional research projects were funded in February 2006. Since its inception MIDAS has been heavily involved in direct support to policy makers, primarily in the area of pandemic flu preparedness. This involvement has substituted for, or been in addition to, work on individual groups' originally proposed aims. The assistance to policy makers was provided both by individual groups and collectively. The consortium has published findings both individually and collaboratively and has submitted one paper as a group.

Soon after MIDAS began, NIGMS staff began developing plans for a formal process evaluation. Abt Associates received funding in 2006, conducted extensive interviews and analyses, and produced a report in late 2007. However, this was not the first examination of the consortium's activities. The role of modeling in pandemic flu planning was the subject of a 2006 National Academy of Sciences (Institute of Medicine) study that was generally favorable to modeling. The report emphasized that modeling is one of many valuable tools available to policy makers, and urged modelers to clearly understand and state the strengths and limitations of their models. In addition to this targeted inquiry, NIGMS staff requested a White Paper from the epidemiological research community that would summarize the status and future needs of infectious disease dynamics and control in general. The then-chair of the MIDAS Steering Committee, Dr. Bryan Grenfell, provided this document to NIGMS Council at its January 2007 meeting. Based on these three overviews and evaluations as well as staff experience with the MIDAS and similar programs, staff is recommending several specific changes to the MIDAS program.

WHAT WE LEARNED FROM THE IOM REPORT AND THE ABT EVALUATION

1. The role of MIDAS participants in providing expert consultation to policymakers

The original (academic) vision of MIDAS as a resource to be tapped in emergencies has proven naïve. Modeling tools are best used for planning and exploring policy options well before there is an emergency. Furthermore, basic science has an uneasy relationship with policy making. Scientists' comfort with uncertainty and their slow, deliberate time frame may be inconsistent with policy makers' need to make good decisions quickly. Policy makers' questions are often specific and detailed; scientists' questions are aimed at general principles. Modelers may not take account of many political realities of policy making.

As our vision has become more sophisticated, we recognize that MIDAS needs to change, both to protect MIDAS investigators' time and energy and to provide useful knowledge and tools to policy makers. MIDAS participants have experienced both the satisfaction of their models being considered by policymakers and the disruption to their research caused by the demands of these consultations. Policy makers have found model results to be informative and useful but are often frustrated when modelers fail to explain their work clearly or to provide explicit answers.

2. The role of the Informatics Group

The informatics group is, according to the original announcement, responsible for collection, curation, and quality control of material in MIDAS database, including data, tools, models, and results. Its primary research component was developing methods to compare and characterize computational models. Based on needs of the research groups, RTI, the current informatics group awardee, had to modify its originally planned role substantially to include providing computational resources, interfacing with supercomputer centers, and creating sophisticated synthetic populations. RTI provided computer resources at critical junctures of the pandemic flu modeling exercise, but this need is being eclipsed as modelers accumulate their own clusters or make use of distributed computing through the Supercomputer Centers. Because RTI had limited access to the PI's modeling algorithms it was unable to statistically compare the algorithms, apply validation techniques, or to provide expert polishing of the underlying code. However, it has been very successful gathering and organizing data through the MIDAS Historical Document and Data Catalog, and using some of the data to develop synthetic populations (U.S. human population, commercial chicken populations) that have direct utility to modelers. RTI also provided expert assistance in developing a global model of pandemic flu emergence and spread, and hardening the code for the TransStat program, and parallelizing code to run on TeraGrid (the NSF supercomputer grid); however, there have been limited collaborations of this type. The informatics group is clearly expert in providing and managing critical data, especially spatial data (through GIS technology), software development, and management of the MIDAS website portal. Given these observations, the role of the Informatics Group needs to be reformulated to closely match the developing needs of the program.

WHAT WE LEARNED FROM THE NIGMS COUNCIL WHITE PAPER

The Future Needs of the Infectious Disease Modeling Community

The NIGMS Council white paper on modeling infectious disease dynamics and control noted that several NIH programs have had a key role in expanding the scope of infectious disease research; these include MIDAS, the FIC/NSF ecology of infectious disease program, and the NIGMS evolution of infectious diseases program. The white paper emphasized that research on modeling infectious diseases is growing in both scope (e.g., behavior and social dynamics, evolutionary processes, and explicit spatial representations) and depth (e.g., the development of statistical and mathematical methods to link in-host and population level dynamics). The paper identified several emerging areas, including vector ecology, seasonal climatic variation, and detailed infection dynamics. Looking ahead, if the field is to advance, it needs

• Infusion of new expertise such as network theorists and social scientists;
• Interdisciplinary training of the next generation of researchers and public health authorities;
• Methods to evaluate models, itself a research question with both basic and policy implications;
• Methods for obtaining, archiving and disseminating epidemiological data;
• Integration of theoretical and field studies;
• Effective strategies for providing the results of research to the public health community.

STRUCTURE AND ORGANIZATION OF A NEW MIDAS PROGRAM

As a result of these considerations, staff proposes that the next generation of MIDAS consist of the following elements:

• A Specialized Center (U54) component to provide leadership in research, training, outreach, and policy;
• A research grant program substantially the same as the existing MIDAS program, to push the envelope of modeling as a tool to analyze infectious disease dynamics; and
• A resource grant (U24) component to address the need for software development; data synthesis, collection, management, and distribution; and specialized services to the research community.

We recommend that MIDAS continue to develop and carry out the program using collaborative projects involving many or all grantees. We also recommend some changes to the organization of the MIDAS steering committee.

MIDAS Centers of Excellence (U54) (RFA)
Proposed: Two to Three 5-year awards
$2 million each (DC)

Currently: no preexisting awards

Consistent with the Institute of Medicine, the white paper, and the MIDAS evaluation we believe that some issues can best be addressed in multidisciplinary centers. We envision supporting three Centers to provide leadership in the following areas:

Science
The goal of MIDAS Centers of Excellence should be to encourage and integrate multidisciplinary approaches to modeling infectious disease dynamics. Centers would be expected to provide national and international leadership. Good modeling rests on a thorough understanding of the system including pathogen and host biology; evolution and ecology; social structure and behavior; and national, state, and local policies and practices—all integrated across space and time. The computational, mathematical, and statistical issues are no less complex. How much detail needs to be in a model? What are the best ways to verify and validate models? Can we devise statistical methods to characterize models and compare their results? And the ubiquitous systems question—how to computationally manage the integration of information across different levels or scales of organization.

Center applications must provide plans for development and dissemination of resources, including data and modeling tools, and they will be required to comply with MIDAS policies. Annual funding will be contingent on compliance.

Policy
MIDAS Centers of Excellence must address narrowing the gap between science and policy related to infectious diseases, including such areas as

• The role of modeling in decision making,
• Fostering discussion among modelers, public health communities, and decision-makers,
• Identifying and addressing the decision-makers needs for modeling results and tools, and
• Ethical considerations in modeling infectious disease policies and practices.

The Centers are expected to work with NIGMS staff to link MIDAS with ongoing and emerging modeling efforts such as those at the Centers for Disease Control and Prevention and at DHHS/OS.

Training and Outreach
To take advantage of the computational, public health, and scientific resources that are available, a new generation of scientists needs to be trained to use modeling tools to study the dynamics of infectious diseases. MIDAS Centers of Excellence would provide cross-disciplinary training in modeling, infectious disease dynamics, public health and policy, and ethics. Centers will also provide opportunities, such as internships, symposia and workshops, for the public health community, decision-makers, and others to study the role of modeling and learn to use modeling tools appropriately and foster dialog between modelers and the public health community.

MIDAS Research Grants (U01): 5 year awards (PAR)
Proposed: Four 5-year awards
$500,000 apiece (DC)

Some investigators can best contribute through regular investigator-driven research grants, unencumbered by the requirements of a Center. A research grant program for modeling infectious diseases would focus on driving forward the field of modeling various aspects of infectious disease dynamics. Grant applications would need to include two components – (1) model development, verification and validation and (2) the science of infectious disease dynamics, which may include estimation of parameters, antimicrobial resistance, infectious disease forensics, social behavior, population dynamics, and ecological and evolutionary questions. Through the cooperative agreeement mechanism, research grant scientists would be expected to participate fully in MIDAS meetings and collaborative projects, and be available for consultations in the event of infectious disease emergencies or anticipation of such.

Research grant applications must provide plans for development and dissemination of resources, including data and modeling tools, and they will be required to comply with MIDAS policies. Annual funding will be contingent on compliance.

All of the current awardees will have to recompete for continued funding under this U01 PAR.

MIDAS Information Technology Resource (U24): 5 year award
Proposed: One 5-year award
$1.5 million (DC)

Currently: One 5-year award (U01, RTI MIDAS Informatics Group), $2.5 million (DC)

Based on experience during the pilot phase of MIDAS, we believe that there are two components of a modeling program that are best handled centrally.

Software Development
Academic computer code may serve its research purpose, but it is not likely to be widely useful due to programming errors, the lack of documentation, and absence of a user interface. Transforming academic code into tools that are useful to the research, public health, and decision-making communities is a job for software professionals and requires considerable government oversight. This function will be episodic, and may need to be supported by occasional supplements.

Data Collection, Management, and Generation
During the pilot phase, MIDAS successfully developed a historical data and document catalog, a detailed catalog of MIDAS models and their results, and a set of synthetic human and animal populations. The latter component requires generating spatial and temporal data that represent (but are not) real demographics and dynamics. We would like to expand data collection, management, and generation to include more infectious diseases, ecological, and climate data, as well as human, agricultural, wildlife, and pathogen information.

Applications for support centers must include plans for development and dissemination of tools, data, and model results.

MIDAS Collaborations
Collaborative projects have been effective in driving the science of modeling infectious diseases, a point made by both the MIDAS evaluation and the Institute of Medicine report. Being involved in common work builds the scientific community, but even more important, it promotes community oversight. We envision identifying one or more collaborative projects that will involve the centers, support group(s), and research projects. Among collaborative projects being considered are

• Model comparisons and validation;
• Dynamics of drug-resistant infections (e.g., MDR-TB or MRSA); and
• Management of vector-borne and zoonotic infectious diseases.

MIDAS Steering Committee
Currently, the MIDAS steering committee is made up of outside experts and the PI's of MIDAS grants. It is problematic for the PI's to be part of a steering committee that advises the institute, a point made by the MIDAS evaluation. A reorganized steering committee should consist of outside experts only and should focus on policy development, oversight, long-term planning, and assessment of progress. The PI's should comprise an Executive Group and be charged with implementation of policies and recommendations and development of MIDAS collaborative projects.

COST

TIMELINE

The first MIDAS U01 awards will expire in May 2009. The succeeding generation of U01 awards will expire in January 2011. Because some of the current U01 grantees may elect to apply for Center grants, the deadline for renewal of their U01's occurs one Council prior to the deadline for Center applications; in the event that a grantee is successful in acquiring a U01 and a Center grant, the U01 will be folded into the Center. The Table tracks the complete turnover of current MIDAS awards.