ESOF2014: From Pathogens to Pandemics: Can we Handle the Risk?

(July 25th, 2014) Wouldn't it be great to be one step ahead of vicious viruses causing influenza, Aids and West Nile fever? Italian physicist, Vittoria Colizza, uses computational epidemiology to understand and stop epidemic outbreaks. She told us all about it during ESOF2014.



Vittoria Colizza studied physics at the University of Rome and received her PhD degree in Statistical and Biological Physics from the University of Trieste in 2004. Since 2011, she has been a Senior Research Scientist at the Inserm, Pierre Louis Institute of Epidemiology and Public Health in Paris, leading the EPIcx, Epidemics in Complex Environments, lab. In her research she characterises and models the spread of emerging infectious diseases with computational approaches. This means, she develops models of a certain pandemic and assesses her predictions in real-time, as for example in the 2009 H1N1 pandemic influenza. This helps to control an acute emerging pandemic and to limit the fatal effects on the human population. At the ESOF 2014 in Copenhagen, she discussed whether we can handle the risk of pandemics and told us about her approach in an exclusive interview:


Lab Times: Why is it so important to model pandemics?

Simple answer: Because we cannot do experiments in real life in order to model how the next pandemic will spread in the world. To better face the pandemics, fight and control it, we have to have some knowledge. Prior to the development of models, people relied on prior pandemics. But clearly, if we have to compare our world in 2009, when the swine flu erupted, against the world of the previous pandemics, which was in 1968, our worlds are totally different in terms of interactions, speed and ease of travel. So, nowadays we have models to reproduce, in silico, what is likely to happen and also test different interventions, compare their efficiency and also costs.


LT: How does a model for predicting epidemics work?

First of all, you have a question in mind; you want to model, for example, the spread of an infectious disease. Then you have to identify the parameters relevant to answer your question. In our case, as we are talking about infectious diseases, which are directly transmitted between humans, we want to have information about humans. We want to know how they get in touch and how they move in space because we are interested in spatial propagation. Then you need to look how to describe these parameters. You need actual data to get a model, which describes what your parameters do. And once you put all these together - it is typically a mathematical and computational model - you run it on the computer and get simulations of the reality. This corresponds to an abstract realisation, where many details are, of course, left out because they are not relevant to your question.


LT: And you can predict the presence and future?

Well, one thing at a time (laughs). First of all, when you have a model, you would like to test it. For example, whether the parameters you put in are really the important ones. Are you able to reproduce what you observe in reality? This is no prediction but it is very important because it allows you to identify the key mechanisms at play. Once you know them, you can also act on those mechanisms, for example to control an epidemic. Then in a second step, you can model different scenarios: What if…? These are hypothetical scenarios in your fake reality, in which you test whether or not your interventions would be efficient. This is the step of preparedness - the more ambitious one is making predictions.


LT: What sources do you use to get your data?

The global model is based on three data layers: One of population and distribution in space and two of mobility. We get data from satellite images, which map population distribution all over the world. Then, we couple this information with mobility data, mainly airline traffic. We have the data from the international air transport association database with around 3,500 airports and 15,000 connections around the world, including information on how many passengers fly on each connection. This is also coupled with the third layer, which is commuting. So, air travel is important for international spread but we also have to think about close cities. It is also important to remember that we are talking of huge datasets and at the same time different time scales: within one day, if we are talking about commuting, and more than one day, for larger spatial scale. All of these have to be integrated into a model, using specific approaches.


LT: What about face to face interactions between people?

For a global model, this data is not integrated at the moment. We learn a lot from smaller scale experiments with RFID [Radio-frequency identification] tags in order to track face-to-face contacts between people, and this can better inform us how to model that. It is another part of the topic and yet another one is using individual data obtained from mobile phones to track individual mobility in space.


LT: Can you test vaccines or medicines?   

Yes, we can test, for example, the implementation of the use of pharmaceuticals like medicines for treatment or vaccines, and also how to distribute them between countries with different resources. We can model different scenarios, in which we could optimise the use of the same amount of resources just by distributing them differently. Here, we could have a big advantage.


LT: How about helping to develop antibiotics?

With these models we can not do this. These models assume as input information on the efficacy of pharmaceuticals, which we get, for example, from clinical trials. We can model the effect of pharmaceuticals on the spread of an pandemic very realistically, but the models cannot help in terms of identifying new drug targets. So, we can only test what is available. But we can test uncertainties, for example, if the drug is less effective than expected, what is going to happen.


LT: Can you predict when and where the next pandemic will start?

Well, this is not something we do; our work starts when a new pandemic emerges.


LT: So, you are happy when a new pandemic breaks out?

Totally not! It is a lot of work besides being unhappy for health reasons (laughs).


LT: Do you think we can handle the risk of pandemics?

Well, there are clearly huge risks and we are trying to tackle them from different angles. Clearly, we know more than what we knew 100 years ago and we have new technologies in many different fields that can help us. But I wouldn't dare to say the war is over - they said this 100 years ago and they were wrong, 50 years ago and they were wrong - so clearly not.

Karin Lauschke

Photo: www.esof2014.org, Vittoria Colizza




Last Changes: 09.09.2014



Information 4


Information 5


Information 6