Recent public health crises, ranging from the 2009 H1N1 influenza pandemic to the Deepwater Horizon oil spill and Fukushima Daiichi nuclear accident required leaders at all levels of government and in the private sector to make decisions during times when knowledge was highly uncertain and limited. At the time when these events emerged, information was confined to limited reporting exchanged primarily through the media and response organizations. But each time, the scientific modeling community quickly engaged with public health response efforts to help consider how these events might unfold using computational models to generate alternative scenarios and predict outcomes. For example, during the 2009 H1N1 pandemic, models were used to forecast the possible number of cases, the implications of interventions such as school-closures and vaccine delivery schedules, and the potential for multiple waves of disease. During the Deepwater Horizon oil spill, models were used to forecast the amount, location, and movement of contaminants and the potential health hazards from exposure to them. Drawing from a wide range of expertise, scientists rapidly assembled a scenario framework to identify alternative futures.1 Following the Fukushima Daiichi accident, scientists produced models to project radiation levels and their implications for populations downwind from the affected area and the potential benefits of medical countermeasures such as potassium iodide.
In each situation, decision-makers used model results to narrow the zone of uncertainty, estimate consequences, and identify potential actions for consideration. They also faced challenges such as interpreting model uncertainties, determining the value of model-based information compared to other sources of predictions including the media and topical experts, and choosing among models and modeling approaches, particularly when the estimates derived from them differed. This process required active collaboration and communication between modelers and public health officials to quickly assess scientific information, evaluate scenarios, and identify possible interventions that could benefit outcomes.
Scientific Preparedness and Response in the Modeling Community
The ability of the scientific modeling community to meaningfully contribute to postevent response activities during public health emergencies was the direct result of a discrete set of preparedness activities as well as advances in theory and technology. Scientists and decision-makers have recognized the value of developing scientific tools (eg, models, data sets, communities of practice) to prepare them to be able to respond quickly—in a manner similar to preparedness activities by first responders and emergency managers. Computational models have matured in their ability to better inform response plans by modeling human behaviors and complex systems. Furthermore, the speed of computing technology has increased dramatically, making it possible to review the outcome from some types of models very quickly. For example, many models used during response activities were developed and validated to advance research on disease transmission, atmospheric dispersion, and ocean circulation long before they were used to inform decisions during a public health response. Systems were already in place to capture and share scientific data necessary to calibrate these models. Research networks such as the Modeling of Infectious Disease Agent Study (MIDAS) program funded by the National Institutes of Health were able to rapidly mobilize research teams to answer scientific questions and were willing to support decision-makers' questions.
Disaster experts recognize that a core set of capabilities is needed to respond to most disasters, and these capabilities are often exercised as a core part of preparedness activities. The modeling community has made tremendous progress in identifying a core set of capabilities—becoming scientifically prepared—to respond rapidly with robust and (hopefully) accurate models that can help policy makers and the public limit adverse outcomes.
We advocate for further development of science preparedness activities as deliberate actions taken in advance of an unpredicted event (or an event with unknown consequences) to increase the scientific tools and evidence base available to decision-makers and the whole-of-community to limit adverse outcomes. Furthermore, science preparedness activities lay the foundation for making scientific research a part of public health emergency response by ensuring that the scientific community can be rapidly mobilized and begin scientific research that is vital to understanding and managing the emergency and preparing for future ones. Prior successes clearly demonstrate the value of modeling to an integrated science response and the importance of developing these capabilities further. Science preparedness and response activities, as we envision them for the modeling community, provide more than rapid production of model results. They also can generate new research questions, allow new communities of scientists to interface on important issues, and tee-up emerging questions for other scientists.
Lurie et al identify numerous organizational and administrative bottlenecks to conducting scientific research during disasters and outline a series of steps for addressing them that are applicable to the broader scientific community.2 These include rostering of experts in key topical research areas with development of an “on-call” cadre of scientists who can be mobilized to help identify scientific priorities and conduct research; preapproved study protocols and survey instruments; mechanisms for rapid human subjects review and approval for postevent studies; and identification of predesignated, prefunded research networks to support focused research studies. Similar activities specific to modeling sciences can also be identified. The Table identifies a list of research goals, preevent activities, and benefits that comprise the beginning of a strategy for science preparedness in the scientific modeling community.
These activities must be in place before an event, so that decision-makers can derive the most value from them, data collection and information sharing efforts can improve the science necessary for many public health responses, and decision-making processes can support the inclusion of actionable model results. These activities will also connect more broadly to other fields that have important scientific roles during a response.
In addition to having models available, it is also imperative that decision-makers are adequately prepared to use models, and that modelers are prepared to work with decision-makers. Science preparedness activities that address this challenge include vetting models in advance and developing processes and frameworks to use them in advance of events. Implementing this entails systematically working through the various types of models, knowing which models are useful in what situation, and establishing the necessary knowledge about models and visibility of models to aid decision-makers. Furthermore, this requires an interface between modeling communities and public health practitioner communities to establish a common dialogue in advance of response and recovery phases. Finally, it also requires a shared level of sophistication and vocabulary to ensure the value and limitations are well-understood and articulated clearly.
Models can be powerful tools to support decision-making when they are used appropriately. They can help decision-makers develop heuristics, identify potential biases, and pay attention to critical system features that play an important role in resiliency. The urgency and rapidity of unfolding public health emergencies may make it challenging to develop precise models in time to inform response options, but understanding system dynamics can lead decision-makers to make better decisions by understanding and exploring the implications of these options.
The Dynamics of Preparedness conference was an excellent example of the type of science activity that can set the stage for improving our nation's preparedness. It brought together leading scientists and public health officials who have real-world experience responding to disasters to discuss new capabilities in mathematical and computational modeling research specific to public health preparedness. What made this conference unique and important was an ongoing discussion of the value of modeling as a scientific capability for the specific purpose of increasing our public health preparedness. Participants identified many advances for rapid situational awareness, forecasting, scenario exploration, and resilience studies—each of which is beneficial within a specific context, for a small amount of time, and with limited data—and identified ways to operationalize modeling as science in a state of readiness to respond.
1. Machlis GE, McNutt MK. Scenario-building for the deepwater horizon oil spill. Science. 2010;329(5995):1018–1019.
2. Lurie N, Manolio T, Patterson A, Collins F, Frieden T. Research as a part of public health emergency response. N Engl J Med. 2013;368:1251–1255.
computational modeling; decision-making; disasters; science preparedness; science response