In Locked Down Spain, a Physicist's Disease Model Research Turns Strikingly Real
Image courtesy of Yamir Moreno.
(Inside Science) -- Yamir Moreno is a physicist at the University of Zaragoza in Spain, where he is also director of the Institute for Biocomputation and Physics of Complex Systems. For about 20 years, he has applied the tools of physics to improve models of disease spread. He was among the pioneering researchers who first incorporated knowledge about humans’ contact networks and everyday movements to make epidemic models more realistic.
Inside Science’s Catherine Meyers talked to Moreno on March 17, shortly after Spain had announced a countrywide lockdown in an effort to slow the spread of the new coronavirus. As of March 20, deaths from the disease in Spain exceed 1,000. The following interview has been edited for brevity and clarity.
Catherine Meyers: How is the mood in Spain right now?
Yamir Moreno: It seems people are taking this seriously. If you go on the streets, it's almost empty, there are only people going for essential work. It's too early to know how this will impact people psychologically. As time goes by, [these measures] could collapse from pressure. It's a complete breakdown of people’s normal routine. That's something we don't know yet. But up to now, people’s mood is okay.
How severe are the restrictions?
They are relatively severe. There is enforced teleworking. The university is completely closed. All restaurants, and places where people can gather are closed. Unless you have to go for work or supplies, or are returning home, you cannot be on the street, either in your car or walking. If you have a dog and you are a couple, only one is allowed to go with the dog outside. But there are still some places that are open, some factories that are still open. There are still some shops that are open, shops where you can buy bread, milk, these kinds of things. Supermarkets are open, of course.
How have you had to adjust your own work?
I am the director of my institute, so I still have to go from time to time to check that everything is okay. We have bio labs, and we are granting some access to a company that purifies proteins that are needed for this COVID-19 test. This is one part of the institute that has to be working. The other part is related to high-performance computing. It's the minimum activity, the essential activity, and as the director of the institute I have to oversee these sorts of things.
From a more personal point of view, I'm working from home. I have some teaching. I have to lecture online whenever possible and have remote interactions with the students.
What sort of research are you doing on the spread of the new coronavirus and how has it changed as the epidemic progressed?
At the very beginning we had time to make predictions, make forecasts. We started to study this in mid-January. But as you enter this explosive phase, where the number of cases grows exponentially, you also need to rush a lot to evaluate scenarios, to try to quantify the facts, the impacts of the measures being taken. For that, we don't usually have a lot of time. There are two approaches -- you can provide rough estimates, things like what you should expect to see if the measures are working. This is one thing.
The other thing is to approach it not for this moment exactly, but as something that is more of a scenario assessment. You try different approaches, even if they will not apply now. This is some knowledge that you will have for the next one. Because there will be a next one.
How might your research or research like it be used by health officials?
A model is a model. You have to be clear with whomever you talk about what your assumptions are, and say, "Okay, under these assumptions, this is what I get." Because sometimes these assumptions are not the same from one day to the next. If this is valid today, the next day it could not be. So, it's important to be honest with ourselves and say, "This is our model, these are the limitations of the model, these are my assumptions." That's okay. Whoever makes the decisions should be in the capacity to judge whether this is important or not.
We provide feedback, put everything on the table; the crisis unit should be able to pick up what is the best solution.
What questions are you looking at right now?
One set of important questions that might have some impact for the future if we move fast enough is: Given that I have adopted these measures, how long can I afford to keep them running? And then how should we, say, reconnect the system? How do I come back to the initial state, where everyone was working, everyone was moving freely? Should I do this slowly, should I do this when I reach a particular level of incidence?
The problem is that we are dealing with humans. Everything is like exploring a landscape of possibilities for which you don't know the probabilities. I have to wait to see which one is more likely to happen, which one is not, and then try to narrow down the set of possibilities. And for that you need time that you don't have, because there are people dying. It's a problem that is really hard. I think no one knows the answer.
Do you have any sense of how long this situation might last globally?
First the peak was in China and Asia. Now it is moving to Europe. I think the peak in most of Europe will be around mid-April or so. Then it will be the Americas, not only the U.S., the whole of the Americas. A question becomes, once you pass the peak, if you open your borders, if you again allow in people from countries where the explosive phase is taking place, what happens?
I think the most optimistic scenario is around two months [for strict social distancing rules]. There are some scenarios in which you could be in a loop that lasts five to six months. We can try to control it, we can try to coexist with this, but the virus will not be under control unless we develop new pharmacological treatments.
With our models, we can advise the authorities how to minimize the impact, but it will not be gone unless new pharmacological interventions are in place.
How did you get interested in epidemic modeling?
My background is in statistical physics. I did my Ph.D. in fracture mechanics -- applying statistical models to the rupture of materials, earthquakes, this kind of thing. When I finished my Ph.D., I wanted to change a little bit, start to do new things. I was lucky enough at that time that this field of network epidemiology was starting. That was in 2000.
I am curious and I like to move beyond the boundaries of different disciplines. Networked epidemiology is one thing that is long-lasting in my research. It is one example in which you can have a real impact in society and human life.
How has the field progressed over the last 20 years?
The basic epidemic models were developed about a century ago by applied mathematicians. The questions we have answered more recently are related to incorporating realism into disease models -- things such as how people move, how they interact, how they are distributed by age, etc. We have also worked on how to deal with systems that are very small, and also systems that are very large. Sometimes what is important for a very small system is not the same as what's important for a very large system. You have to ask: What is the right scale to describe your model? Do you go deep into the details, or just work at the more macroscopic level? We have been successful, more or less, in these kinds of things.
How do you think the current crisis could change the field of disease modeling?
I hope this helps us realize a couple of things. One is that we need more funds to do research in this field. Up until this point, if I talked to funders and said I need more money to make realistic models to understand the spread of disease, they were not really aware of how important this is. I think that that will change. The European Union is mobilizing resources for research now. You are happy if this happens during a crisis, but you would like to do this not in a crisis, but beforehand.
What are some open questions in the field?
One of the open questions -- and it's obvious that it's really, really relevant right now -- is how to incorporate changing human behavior into these sorts of models. There are a lot of things to be done there still. How can we incorporate the fact that humans react to the spread of the disease? We don't know how to embed this change into an equation.
For example, you may assume people will change their behavior proportional to the number infected, to the prevalence of the disease -- so the more prevalent, the more tendency they have to stop from traveling, say. We'll have to start incorporating these sorts of things into our models. It will start in a naive way, in the sense that there are no experiments to back it up. In that sense, it is a very hard question. To answer, you need multidisciplinary science, you need to work with social scientists, to work with psychologists, with economists.
We also need data. With this crisis, there are some companies that have been developing apps that, with people's authorization, track their movements. You can try to learn from data like that, and then incorporate it in a model.
At this stage, I think whatever is not based in data, it's as right as whatever you can say, or as wrong as whatever you can say. It's something that you don't really know, and you don't have ground truth to compare it to.