Preparing for the next pandemic takes more than better models: expert consensus recommendations for infectious disease modelling

When a new disease begins to spread, governments need to know how fast it will move, where it will strike next, and which actions will stop it. To answer these questions, scientists build mathematical models. These are not crystal balls that predict the future with certainty; rather, they are structured ways of thinking about how a virus might behave under different conditions. Imagine a model as a complex map that shows possible routes a traveler could take, based on the roads available and the weather expected. If the map is drawn with poor data or hidden assumptions, the traveler might take a wrong turn. For decades, these models have helped officials manage outbreaks, from foot-and-mouth disease in cattle to the global spread of influenza. However, the recent pandemic revealed that having better maps is not enough. The real challenge lies in how the maps are drawn, how the data feeding them is shared, and how the results are explained to the people who must make life-or-death decisions.

A large group of experts has now come together to define exactly how infectious disease modeling should work to be most useful. Led by researchers from the University of Luebeck, this team brought together 41 specialists in mathematics, public health, and social sciences from universities and research institutes across the globe. They did not simply argue about which equations were best. Instead, they conducted a rigorous process to find agreement on the rules of the game. First, they reviewed hundreds of past studies to see what problems scientists had faced before. Then, they held a series of four rounds of discussion and voting, refining their ideas until everyone agreed on a final set of forty recommendations. The goal was to create a shared standard that would make modeling more reliable, transparent, and ready for the next emergency.

The most striking finding of this work is that the experts agreed almost perfectly on forty specific statements. Thirty-eight of these received the highest possible rating, meaning the entire panel saw them as essential. However, the agreement was not about how to build a specific type of model. Instead, the strongest consensus focused on the conditions surrounding the model: the data it uses, the way it is reported, and the people who work with it. The panel found that a modeling group cannot fix these problems alone. While scientists can decide how to write their code or how to explain their uncertainty, they cannot create public data systems, negotiate international data-sharing treaties, or secure long-term funding on their own. These tasks require governments, funding agencies, and policy advisors to act between pandemics, not just during them.

The experts identified that the biggest hurdles are often practical rather than mathematical. For a model to be accurate, it needs timely data that is easy to access and clearly labeled with its limitations. Currently, data often arrives late, is locked behind complicated barriers, or comes in formats that computers cannot read automatically. The panel insisted that data should be available through open, machine-readable systems with clear documentation. They also stressed the need for cross-border agreements so that data can flow freely between countries during a crisis. Without these foundations, even the most sophisticated model will struggle to provide useful guidance. The group also emphasized that models must be transparent. This means scientists should clearly state what assumptions they made, share their code and data so others can check their work, and explain the limits of their predictions.

Another critical point of agreement was how to talk about uncertainty. In the past, models were sometimes treated as oracles that could predict the exact future. The panel clarified that models are tools for exploring possibilities, not for foretelling destiny. A model might show that a virus could spread rapidly if no action is taken, but if officials implement strict measures, that rapid spread might not happen. This does not mean the model failed; it means the model successfully explored a scenario that was avoided. The experts argued that scientists must clearly distinguish between a forecast of what will happen and a scenario of what could happen under specific conditions. They also agreed that uncertainty should be communicated in ranges, showing a spectrum of possible outcomes, rather than as a single, precise number that gives a false sense of accuracy.

The study also highlighted that the people building these models often work in isolation from the people who use them. The panel recommended creating permanent roles that bridge this gap, where experts can translate complex model results into clear advice for policymakers. They also noted that models often overlook vulnerable groups or fail to account for how human behavior changes during a crisis. To fix this, the experts called for models that include diverse populations and for the integration of social science insights. However, they acknowledged that adding too much complexity can make a model hard to use, so the right balance must be found based on the specific question being asked.

Despite the high level of agreement, the panel recognized significant limitations in their work. The experts were drawn almost entirely from high-income countries and academic settings. This means their recommendations reflect the resources and structures available in those regions, which may not exist in low- and middle-income countries where the next pandemic might strike. The group admitted that their list of priorities assumes a functioning government and health system, which is not always the case. They also noted that their study did not fully address the political realities that often prevent good advice from being followed, nor did it deeply explore the economic costs of different interventions.

Ultimately, the paper concludes that preparing for the next pandemic requires more than just better mathematics. It requires building the institutions and infrastructure that allow models to work effectively. The experts offered ten specific priorities for the time between outbreaks, such as establishing interoperable data systems, creating standing interdisciplinary teams, and securing sustained funding. They argued that waiting until a crisis begins to fix these problems is too late. The recommendations are presented not as a final solution, but as a starting point for a global conversation. By agreeing on these standards now, the scientific community hopes to ensure that when the next threat emerges, the tools available to fight it are robust, clear, and ready to serve the public.