The National Institute of General Medical Sciences (NIGMS), a part of the US National Institutes of Health (NIH), was established in 1962 via an Act of Congress for the “conduct and support of research and research training in the general or basic medical sciences and related natural or behavioral sciences”, especially in areas which are interdisciplinary for other institutes under the Act, or alternatively, which fall under no institute’s purview. In these 52 years, the NIGMS has acquitted itself laudably as one of premier funding agencies that support basic research into understanding biological processes, disease diagnostics, treatment and prevention. At any given time, NIGMS supports close to 5000 research grants, accounting for more than 1 in every 10 grants funded by NIH as a whole, and has the distinction of funding the Nobel Prize-winning research of 75 scientists.

In the aftermath of 9/11, the adequacy of the preparedness of the United States against bioterrorism and emergence of new or more virulent microbial pathogens was called into question. The existing biodefence research plan formulated by the National Institute of Allergy and Infectious Disease, the NIH institute dealing directly with microbial pathogens, was well-developed and elaborate, covering important target areas such as microbial biology, host immunity, vaccine prophylaxis, therapeutics and diagnostics. However, its scope did not cover computational or mathematical modeling – a system of collection and analysis of data from real-life scenarios, and development of hypotheses-driven contingency models, which would inform policy decisions. By the December of 2001, concerned policymakers evinced interest in the utility of good scientific models which could do that in response to a bioterrorism threat or an emerging epidemic, both considered variants of the infectious disease transmission process.

NIGMS, with its rich history of supporting computational approaches to complex systems, was naturally the best choice to help add a mathematical modeling component to biodefence research. The strength of such modeling lies in its ability to integrate large amounts of data into the wider context of a complex system, akin to the pieces of a jigsaw puzzle. Models can predict how the parts of the complex system work individually and inter-dependently, and generate plausible hypotheses to account for gaps where the data are inadequate – just as a partly-revealed image on the already locked-in puzzle pieces can provide clues about the whole image and also where the next piece might fit. Aided by powerful computers, these models can also simulate problems in the system, which help detect possible vulnerabilities.

The infectious disease transmission process is complex enough to benefit from such scientific models; the process is influenced by the biology, ecology, natural history and evolution of the microbial component, biology and genetics of the host that determines its response to the microbial challenge, as well as interactions with the external environment which include prophylactic and therapeutic intervention strategies – all of which a good model would have to account for. Microbial ecology, which includes undiscovered reservoirs of the microbe in the wild as well as unidentified modes of transmission, assumes special significance for emerging pathogens, which may infect and inhabit multi-species hosts, either causing disease or being carried passively. Therefore, ecology, as well as the steadily growing understanding of microbial genomics, continue to serve as important data sources feeding into the computational models.

With this august goal of understanding disease spread and impact, NIGMS in 2003 introduced MIDAS – the Models of Infectious Disease Agent Study – its flagship program for building predictive models of infectious disease outbreaks based on mathematics and statistics applied to real-life data.

Infectious Disease Modeling

Virtual simulations performed by MIDAS researchers may sound like an absorbing computer game. Using population demographic and geographical data, the researchers build virtual communities, urban or rural, incorporating different environments, such as home, public amenities, schools, places of work and business, and so forth. Within these virtual communities, the researchers simulate the release and spread of an infectious agent, calculated based on real-life epidemiological data and actual/estimated information about community habits.

These models can be tweaked appropriately to ask different questions, such as the dynamics of pathogen spread, virulence of the pathogen and how it affects host-pathogen interaction, effectiveness of intervention strategies, and so forth. The strong computational support also comes in handy to identify disease agents and conduct disease surveillance – making it possible to visualize different emergent situations, from random infectious disease epidemics to deliberate acts of bioterrorism. Understandably, this process requires input from an interdisciplinary team comprised of infectious disease professionals and epidemiologists, statisticians, bioinformatics and computational biology experts, as well as social scientists.

See a quick explanation of how the modeling is done, in the context of malaria research; this video is courtesy the Intellectual Ventures Laboratory.

The risk- and outcome-analyses arrived at via these computational models provide a much more comprehensive and quicker understanding of disease spread than would be possible from existing, human communication-based methods. This makes the MIDAS program invaluable for planning policy for a swift and appropriate response to disease outbreaks. To achieve this, MIDAS researchers actively collaborate with local public health officials and the US Centers for Disease Control and Prevention (CDC), as well as internationally with the World Health Organization (WHO) and the European Centers for Disease Control. MIDAS results and resources are also shared with the wider scientific and public health community, following NIH guidelines, via journal publications, presentations, and most importantly, the MIDAS portal.

Advancements and Accomplishments

Just in its 11th year, MIDAS already boasts of an impressive range of accomplishments via its constituent research centers (referred to as MIDAS National Centers of Excellence). Some noteworthy examples include the following:

FRED (Framework for Reconstructing Epidemiological Dynamics) is a modeling software developed by the University of Pittsburgh Center in collaboration with Carnegie Mellon University, with financial support from NIGMS/MIDAS and the Bill and Melinda Gates Foundation; it uses virtual (a.k.a. ‘synthetic’) populations mentioned above, created from census-based data and social networks, in order to study viral disease outbreaks. FRED can simulate viral multiplication and evolution in the wild, and estimate the effects of control strategies such as vaccination and anti-viral medication, as well as public health measures, such as separation of vulnerable populations (via school closures) and promotion of beneficial personal health behaviors, hygiene and vaccine acceptance (1). This center has used FRED to model the 2009 H1N1 influenza epidemic simulating the conditions in different cities (2, 3), and predicting the benefits of vaccination (4-7) – as well as making policy recommendations (2, 4, 5, 8, 9) to combat the next potential outbreak. Dr. Irene Eckstrand, a program director at NIGMS, opines that FRED, which could run the simulations at the backend and deliver the results to a smart device, could potentially enable public health officials to employ modeling tools away from the desk.

In addition to FRED, researchers from Pittsburgh and other centers have developed a host of tools dedicated to epidemiological modeling and research, such as:

  • Geospatial Area and Information Analyzer (GAIA), a visualization and analysis tool for epidemiological research;
  • Highly Extensible Resource for Modeling Event-Driven Supply Chains (HERMES), a tool for optimizing supply chains;
  • Indicators for Stress-Adaptation Analytics (ISAAC), a data source for modeling public health system operations;
  • Legal Networks Analyzer (LENA), a tool to improve emergency preparedness and response capacity in state and local public health systems, which allows policymakers to visualize legally directed relationships between public health system agents as mandated by federal or state statutes and regulations;
  • EpiFire, a programming interface for simulating the spread of epidemics on contact networks;
  • Global Epidemic Model, a tool to study the effects of global air travel on the spread of infectious diseases;
  • FluTE, a stochastic influenza epidemic simulation model; and
  • TranStat, an analytic tool for infectious disease outbreak data that models host to host transmission and epidemiology.

A more complete list along with source links is available at the website for MIDAS models and resources.

The MIDAS program at Harvard School of Public Health is run under the Center for Communicable Disease Dynamics, which is engaged in mathematical modeling and statistical analysis of infectious disease data to explore various aspects of the dynamics of disease transmission (including disease emergence, seasonality, interactions between diseases, and so forth) as well as epidemiological surveillance to aid decision making; the program also aims to develop early detection and tracking models for infectious disease-associated phenomena, such as outbreaks (10) and emergence of drug resistance (11). Already in progress is work directed towards anti-retroviral therapy (12), influenza preparedness (13), malaria control (14), cholera outbreak (15), effectiveness of pneumococcal vaccination (16), and drug resistant tuberculosis, among others.

Other collaborative projects of note include the NGIMS-funded MIDAS programs at:

(a) the University of Chicago and Argonne National Laboratory, which uses urban modeling studies to understand the spread of a serious public health threat, MRSA (methicillin-resistant Staphylococcus aureus, a drug-resistant bacterium) within certain communities (17);

(b) the University of Washington –  Fred Hutchison Cancer Research Center, which employs mathematical modeling and virtual epidemic simulations to study transmission of disease agents, whether naturally occurring (outbreaks) or maliciously spread (bioterrorism), in order to come up with best intervention strategies, including vaccination and antimicrobial prophylaxis, workable even in resource-restricted scenarios (18); this group recently received an NIGMS award to model and track various pathogens, including Ebola virus; and

(c) the Virginia Bioinformatics Institute of Virginia Tech, whose active modeling system can track the transmission of simultaneously circulating diseases in a simulated urban population created based on demographic and geographic data; this group (along with researchers from Fred Hutchison) recently provided an early warning about extremely high numbers of Ebola cases in West Africa, arrived at via predictive, adaptive modeling.

The centers for excellence under the MIDAS program also participate in outreach activities; for instance, funded by various US government and private sector organizations, the Virginia Bioinformatics Institute runs initiatives that take computational modeling and bioinformatics to K-12 students, including STEM summer programs and computer games that teach how viral diseases spread.

The MIDAS centers for excellence are supported by Research Triangle Institute, a MIDAS information technology resource group; in addition to the supportive duties, RTI maintains information repositories pertaining to various disease outbreaks, and they have also developed detailed virtual human and livestock populations for United States using census, other demographic and geographical data. Similar computer-simulations and predictive models, developed by MIDAS centers to study epidemiological and social aspects of disease spread and intervention strategies, have been employed for Colombia (19), Chile (20), Mexico (21), Thailand (22), and China (23).

Impact on People’s Lives

Computational or mathematical modeling work is perhaps not the most glamorous of scientific jobs, and is certainly not the most visible. Whenever emerging diseases, epidemics and other events associated with public health are concerned in real life, people tend to think in terms of direct action – physicians and laboratory personnel for diagnosis and treatment, public health officials for strategy decisions related to prevention and interventions, and even lawmakers for official promulgation of policy and provision of funding for relevant activities. However, a nation’s state of preparedness for potential public health disasters, such as an infectious disease outbreak, depends on a lot more – much of which is required to be done in the background, without fanfare.

This is where bioinformaticians and computational scientists come in; their role is to: (a) aggregate knowledge from various fields, such as medicine, microbiology, and immunology; (b) collect and gather data from epidemiological and disease-surveillance studies; and (c) analyze the data to generate models, or digital representations of what happens in a real-life situation. This process requires powerful computers guided by smart software, because the computer processor is able to handle the multitude of parallel calculations that are needed for an efficient model. A useful model must take into consideration myriads of factors, small and large – such as, (a) the ability of the infectious agent to cause disease (a.k.a. ‘virulence’) and the mechanism of disease (a.k.a. ‘pathogenesis’), (b) human behavior, in person and in a community of people, including interactions and random acts, (c) impact of interventions that medical professionals may undertake, and so forth. Once a model is made, it is tested – and tweaked if necessary – by asking it various questions and presenting different scenarios to its simulated, self-contained ecosystem.

As indicated by the work presented in the references, the MIDAS researchers have repeatedly proven (and continue to prove) the immense value of this program. In 2013, researchers at the University of Pittsburgh made available a digitized database of US disease surveillance weekly reports collected since 1888, which includes nearly 90 million cases related to 56 types of infectious disease; this is an invaluable tool for epidemiologic analysis and modeling, and can, for example, demonstrate the public health benefits reaped from interventions such as vaccination for reducing the tremendous burden of deadly infectious diseases, including smallpox, polio, measles, and whooping cough. Modeling and simulations have made it possible to forecast the initiation, peak and ebb of the influenza season, which aids policy-making and disease preparedness (24); it proved immensely useful during the spread of H1N1 flu and informed the efforts during subsequent outbreaks with newer viral strains. Modeling of Tsetse fly distribution in Kenya identified hotspots for successful eradication of this agent that spreads the deadly disease called sleeping sickness in Africa. More recently, modeling systems are being used to track the spread of antibiotic resistance in disease-causing bacteria; these efforts have borne fruit in identifying the mechanisms by which community-acquired MRSA spread within populations (17) and finding how vancomycin-resistant enterococci from geographically-close hospitals are related (25). As mentioned above, modeling efforts from some groups are currently focused on the spread of the Ebola epidemic, which has been predicted to leave West Africa and spread to previously unaffected parts of Africa and other nations.

Can MIDAS predict everything? Sadly, no. Models such as these are unable to provide exact and accurate numbers in real time for people and societies affected by an epidemic. But based on historical data and analysis of prior patterns, they can certain inform policy with suggestions about interventions which may work. MIDAS models also actively study how disease agents evolve and interact with the environment, which includes the question of seasonality of certain diseases. With these capabilities, MIDAS aims to become a common point of engagement amongst policymakers, decision-makers, public health officials, as well as the general public, offering smart, empirically validated, and reliable guidance over time – the foremost jewel in the crown of NIGMS.


P.S. If you have a half hour to spare, do check out this lecture on Infectious Disease modeling by Marc Lipsitch, Director of the CCDD of the Harvard School of Public Health, and Principal Investigator of the Harvard MIDAS program.

Bibliography

  1. Grefenstette JJ, et al. FRED (a Framework for Reconstructing Epidemic Dynamics): an open-source software system for modeling infectious diseases and control strategies using census-based populations. BMC Public Health. 2013;13:940.
  2. Brown ST, et al. Would school closure for the 2009 H1N1 influenza epidemic have been worth the cost?: a computational simulation of Pennsylvania. BMC Public Health. 2011;11:353.
  3. Cooley P, et al. The role of subway travel in an influenza epidemic: a New York City simulation. J Urban Health. 2011 Oct;88(5):982-95.
  4. Lee BY, et al. A computer simulation of vaccine prioritization, allocation, and rationing during the 2009 H1N1 influenza pandemic. Vaccine. 2010 Jul 12;28(31):4875-9.
  5. Lee BY, et al. A computer simulation of employee vaccination to mitigate an influenza epidemic. Am J Prev Med. 2010 Mar;38(3):247-57.
  6. Lee BY, et al. Vaccination deep into a pandemic wave potential mechanisms for a “third wave” and the impact of vaccination. Am J Prev Med. 2010 Nov;39(5):e21-9.
  7. Lee BY, et al. The benefits to all of ensuring equal and timely access to influenza vaccines in poor communities. Health Aff (Millwood). 2011 Jun;30(6):1141-50.
  8. Kumar S, et al. Policies to reduce influenza in the workplace: impact assessments using an agent-based model. Am J Public Health. 2013 Aug;103(8):1406-11.
  9. Lee BY, et al. Simulating school closure strategies to mitigate an influenza epidemic. J Public Health Manag Pract. 2010 May-Jun;16(3):252-61.
  10. Miller JC, Volz EM. Model hierarchies in edge-based compartmental modeling for infectious disease spread. J Math Biol. 2013 Oct;67(4):869-99.
  11. Ford CB, et al. Mycobacterium tuberculosis mutation rate estimates from different lineages predict substantial differences in the emergence of drug-resistant tuberculosis. Nat Genet. 2013 Jul;45(7):784-90.
  12. Eaton JW, et al. Health benefits, costs, and cost-effectiveness of earlier eligibility for adult antiretroviral therapy and expanded treatment coverage: a combined analysis of 12 mathematical models. Lancet Glob Health. 2014 Jan;2(1):e23-34.
  13. Shaman J, et al. Real-time influenza forecasts during the 2012-2013 season. Nat Commun. 2013;4:2837.
  14. Hansen E, Buckee CO. Modeling the human infectious reservoir for malaria control: does heterogeneity matter? Trends Parasitol. 2013 Jun;29(6):270-5.
  15. Grad YH, et al. Cholera modeling: challenges to quantitative analysis and predicting the impact of interventions. Epidemiology. 2012 Jul;23(4):523-30.
  16. Weinberger DM, et al. Using pneumococcal carriage data to monitor postvaccination changes in invasive disease. Am J Epidemiol. 2013 Nov 1;178(9):1488-95.
  17. Macal CM, et al. Modeling the transmission of community-associated methicillin-resistant Staphylococcus aureus: a dynamic agent-based simulation. J Transl Med. 2014;12:124.
  18. Halloran ME, Longini IM, Jr. Emerging, evolving, and established infectious diseases and interventions. Science. 2014 Sep 12;345(6202):1292-4.
  19. Lee BY, et al. The potential economic value of a Trypanosoma cruzi (Chagas disease) vaccine in Latin America. PLoS Negl Trop Dis. 2010;4(12):e916.
  20. Nsoesie EO, et al. Modeling to predict cases of hantavirus pulmonary syndrome in Chile. PLoS Negl Trop Dis. 2014 Apr;8(4):e2779.
  21. Lipsitch M, et al. Use of cumulative incidence of novel influenza A/H1N1 in foreign travelers to estimate lower bounds on cumulative incidence in Mexico. PLoS One. 2009;4(9):e6895.
  22. Lee BY, et al. Replacing the measles ten-dose vaccine presentation with the single-dose presentation in Thailand. Vaccine. 2011 May 12;29(21):3811-7.
  23. Yu H, et al. Transmission dynamics, border entry screening, and school holidays during the 2009 influenza A (H1N1) pandemic, China. Emerg Infect Dis. 2012 May;18(5):758-66.
  24. Halloran ME, et al. Modeling targeted layered containment of an influenza pandemic in the United States. Proc Natl Acad Sci U S A. 2008 Mar 25;105(12):4639-44.
  25. Lee BY, et al. Modeling the regional spread and control of vancomycin-resistant enterococci. Am J Infect Control. 2013 Aug;41(8):668-73.