From surveillance to preparedness: using AI to stay ahead of outbreaks

Pandemic preparedness is fundamentally about being able to act before an outbreak escalates. The earlier changes in disease transmission can be recognised, the more time public-health teams have to investigate, prepare and respond.

At EPCON, we are using AI to contribute to this shift from reactive surveillance towards earlier, more forward-looking epidemic intelligence. Rather than focusing only on where cases are being reported today, AI can help analyse how disease patterns evolve across locations and over time and identify where risk may be changing next.

Our work with Ebola provides a concrete example. Using a Bayesian AI approach, we combine historical outbreak data with contextual information to estimate how transmission risk may develop from one week to the next. The model considers recent disease activity within a health zone, transmission in neighbouring areas and contextual factors such as population density, access to healthcare, vaccination coverage and water access.

By bringing these signals together, the model can estimate the probability of different levels of Ebola activity in the following week. In our latest model development, this approach achieved approximately 73% accuracy in predicting the following week’s level of Ebola activity, showing the potential of AI to identify meaningful patterns in how transmission risk evolves over time and space. This moves the analysis beyond simply mapping existing cases towards identifying areas where increased transmission may be emerging.

For public-health teams, even a small increase in this window for action can be valuable. Earlier signals could support more targeted surveillance, faster investigation, stronger diagnostic readiness and better preparation or positioning of resources before transmission expands further.

While Ebola is our current example, the wider opportunity is much broader. Different diseases have different transmission dynamics and require different data, but the underlying principle remains highly relevant for pandemic preparedness: combine what we know about previous transmission with what we know about the local context to anticipate how risk may evolve.

This type of approach could support preparedness for a range of epidemic threats by helping decision-makers move from asking only “Where is disease occurring now?” towards also asking “Where might we need to be ready next?”

At EPCON, we see AI as a tool to strengthen, rather than replace, epidemiological expertise. Its value lies in bringing together complex and continuously evolving information and turning it into an additional layer of intelligence for public-health decision-making.

Ultimately, the potential of AI in pandemic preparedness is not prediction for prediction's sake. It is about creating more time to act - detecting changing risk earlier, responding faster and strengthening our ability to contain emerging outbreaks before they grow.