Coronavirus | Densely-packed areas in cities are vulnerable, says biocomplexity expert Madhav Marathe

Madhav Marathe. Photo: Special arrangement  

Madhav Marathe has extensive experience in modelling and studying infectious diseases. His group has supported almost every outbreak response in the US since 2002, including planning for the H5N1 in 2005, H1N1 in 2009, MERS in 2012, Ebola in 2014 and Zika in 2015. In this interview he describes the challenges of responding to the COVID-19 pandemic and discusses India’s potential response. He is an endowed Distinguished Professor in Biocomplexity, Director of the Network Systems Science and Advanced Computing (NSSAC) Division, Biocomplexity Institute and Initiative at the University of Virginia, and an alumnus of IIT Madras.

What is the challenge in modelling or studying a pandemic, COVID-19 in particular?

Covid-19 presents a number of new challenges in my opinion. In this case, three things happened: (1) The pandemic started in a densely populated region of the world unlike Ebola, which started near Liberia in West Africa; (2) Unlike West Africa, China is much more strongly connected with the world, in terms of economic activity and human mobility; and (3) the incubation period of the disease is longer than many other flu viruses. Moreover, a large number of people – the silent spreaders – don’t show any symptoms at all while being infectious. This makes it hard to track the spread of the disease . Finally, the social and economic impact of this pandemic will go way beyond any disaster we have seen in a long time. Recovering from this impact will require a coordinated global response.

What does mathematics tell us about pandemic? Some people are of the opinion we need to keep this lockdown for a year to contain COVID-19?

Here is my take. This is open to scientific debate, so one cannot answer these things with certainty. However, this part, I think, everyone agrees – this is a new virus as far as humans are concerned, hence, everybody is susceptible to the virus. The general view [for the lockdown period] is that until herd immunity sets in, which means until enough number of people have already been infected and recovered, or until you have a vaccine (or an antiviral), this disease will keep circulating in the human population.

One way to speed this up is to let the epidemic go completely unchecked but the downside will be many deaths and hospitalisations – as can be seen in China, Italy and even the U.S., whose numbers are rising fast. So this theoretical option is simply not a possibility. The other option of completely shutting down the society for one year is also not a possibility. What we have at hand is a complex multicriteria optimisation problem.

You can control COVID-19 via social distancing. You have to decide how effective that mechanism should be. We try and buy time until we have antivirals or a vaccine. This reduces the mortality and the number of individuals who get critically ill, thereby reducing the burden of the disease on the healthcare system to a manageable level. So the question then is: (i) what sort of social distancing would you put in place, (ii) when would it begin , and (iii) for how long? Each country and its citizens need to decide what the best solution for them is, with two caveats: (i) human life has to be central in these analysis, (ii) one has to understand that we are globally connected so infections in one part of the world will likely have impacts on other parts of the world. You have also seen very different numbers for different regions – in Italy, we have seen approximately 7,500 deaths till now with 74,000 infections. In Germany, the numbers are remarkably different – we have 35,000 infections, and only 185 deaths. Some countries have managed this better.

Perhaps it is not a question of managing the response better but just the way it started, maybe the demographics, maybe they have more immune-compromised individuals maybe the way testing was done or perhaps the way interventions were implemented. Research will have to be carried out to sort this out.

Different countries are responding differently to this virus. So when you are trying to understand it as a scientist, what would you say are the key differences in modelling it?

The way we address how to understand the spread is to build the digital twin of the city. If you want to understand infectious diseases in a large city such as Chennai, then you want to build a realistic social contact network that represents Chennai. In the social network each node is a person and the edge between each person reflects their social connectivity. For example if you and your neighbour come into contact with each other, there will be an edge between you and the neighbour. The problem is the difficulty in constructing these network in the first place. Especially when studying social networks for a large city like Chennai; once the number is more than 15 or 20 nodes, it becomes very hard to synthesise such networks.

We came up with a novel way of constructing a synthetic representation of this network. i.e. a network representation that statistically captures the interactions but need not necessarily be identical to the population. So, what does it mean to build a “digital twin” or a synthetic city? The idea is, in order to build a social contact network of Chennai, I can take all sorts of information, from census pollings, to roads and buildings, to surveys of what people are doing all day, travel schedules, transit availability etc. and build a network which statistically resembles Chennai. But if you look at any one person in particular you cannot find an identical agent, say, yourself, or my professor at IIT Madras. It is a digital replica but it is a synthesised. That is the network we use to study how diseases would spread using simulations and computer models. We have developed methods that can be run on big supercomputers since the methods are computationally expensive. You look at outcomes by running the simulations of disease progression on these networks. Machine learning techniques and AI tools are used to glean insights from these simulations. This includes forecasting, resource allocation, interventional analysis and so on. This way you can capture the social context which is unique for each city. Network for Chennai is going to be very different than the social network for Delhi, or for Washington DC. Their demographics are different, the population density is different, the role of transportation is different – these things will finally affect the digital twin. It is in that sense that we are able to understand how disease would spread from one region to another, or from one city to another.

In fact, my colleagues have shown in a paper that this social connectivity affects how intervention works. If you had the same set of interventions in two different cities they need not work the same way at all.

You say that different methods for halting the spread may be needed for different cities. Can you give some examples?

First, for this pandemic, there is only one broad-based method – there are no pharmaceuticals at all. The only real method to control this pandemic is social distancing. On the other hand, each region or city or community is different. Intuitively, I would say a city that is much denser would require more stringent social distancing measures than the one that is very sparse.

If you take New York City or big cities in India, public transport is the fundamental mode to move around, but smaller cities have very little public transportation. That has an effect because cutting the public transportation down also means cutting the movement or intervention. But shutting down the public transport system in Houston, where it is limited, is not likely to have that much of an impact as shutting it down in New York.

Another example is that some cities might have a lot of people who are elderly. Usually the southern part of this country (US) has a lot of people who are elderly. In India, it is not common to have large sections of the town devoted only to elderly folks. So, cordoning off communities that house elderly people in the US is a different idea than doing it in India.

What are your ideas for India in the case of this pandemic?

I remain anxious about India, although the steps the Indian government has taken so far are very good. I was very happy to hear that they locked down the entire country; they are trying very hard to control it. But the question is how big will the outbreak in India grow to? If it stays at a low level then we are in good shape. It might be that even though people are getting sick there, the death rate is not as high as in Italy, which would be good news. There could be multiple reasons. For instance, the immune system of Indians could work differently from those of Italians. The government is instituting fairly strong measures – they have learnt from the events so far and have instituted the interventions early. This is a challenge for the government, too. You can shut down all the transportation and have general curfew, but for how long can you do it?

The economic engine of the country essentially comes to a grinding halt.

Additionally, I have seen in India that people often do not comply with government orders. But, in times of crises like this, I feel Indians show an amazing level of awareness and compliance. This is what’s amazing about the Indian society. Our leaders, celebrities, administrators all have to work collectively and constantly remind the citizens of downside if they do not follow the orders. I think the next 10 days for India is very critical. If we are able to slow the disease down and keep it at a very low level, then we have a chance to respond to it better. If numbers start growing exponentially, like they did in Italy or US then controlling it will be very hard, because of the density of the population. We also have slum areas in larger cities like Mumbai, Delhi, Chennai, which are very vulnerable. These are very, very densely packed. People talk about social distancing – it is not easy to define social distancing in a tightly packed slum areas. In some places there is no distance between your house and your neighbours’ house. It is one big house effectively. We need to make sure that these slums are taken care of.

They also lack livelihoods. It is easy to say, “don’t go to work,” but many of them are daily wage earners. It is hard to ask them to stay at home for months when they don’t even have the means to feed themselves. So, if you have strong social distancing measures, the government also needs to think about how it’s going to take care of all of its people. Furthermore, the lockdown implied that many workers left cities for their villages and hometowns. I suspect many of them are already infected. They also travelled in crowded transportation system. This can potentially have implications, we will know this in the next few weeks.

It is also important to understand that shutting down public transportation, might be a good idea, but also restricts people’s movement immensely. So, I think it is challenging in India, but so far the country, the government, and the citizens have shown strong will. Again, though testing has been ramped up, based on my interactions with scientists from India, I feel India has to do more testing. The more you test, the better you can understand how widespread the disease is as well as give some insight into what the mortality rate is.

Testing is non-trivial, the kits are expensive by Indian standards, some $30 a kit. But I am told many companies have started building these kits for testing so that they can do a high level of testing.

The last part is that if it becomes a pandemic the way it has become in the US and people are critically ill in India, the medical establishment will be stretched thin. The doctors, staff, nurses are all very vulnerable without proper protective gear. Not just [in number of] doctors, nurses, and staff who are very vulnerable but whether there are enough ventilators, beds and PPE. Yesterday, on TV, I saw a hospital in Spain where people were on the floor, they did not have enough space. This is Spain, which is very well developed. In India, this kind of situation is likely to happen if the numbers start exploding. The government needs to plan on emergency treatment places. Tent hospitals can treat a large number of folks. China showed how to build such tent hospitals very quickly. India must start planning now. It might seem like a waste, but it is better to plan now. From an operational perspective I have five things that I would like to suggest: (i) Test as much as you can, it is the best investment you can make, (ii) start building hospitals and producing medical equipment and supplies, (iii) be transparent while communicating the extent of the disease; (iv) take care of old sections of the society, especially the poor and vulnerable, and (v) don’t let the guard down even if you dodge the pandemic in the first wave, it will come back.

Is there a strong reason and strategy for testing?

The fundamental reason to test is twofold. One is to identify all the possible infected folks but the other related reason is to essentially estimate the prevalence of the disease. Right now, it is not easy to tell, even in the US, (the exception is Korea where testing was intensive), to even have a sense for how many people are currently infected – we don’t know that in the US. Some people say it is 1%.

In New York, one estimate is that one in thousand people are already infected. They might not be infectious or may not show symptoms, they may not be critically ill, but that is the number estimated. It is a very weak estimate.

But randomised testing, if you had that many kits, would have addressed this better. That also gives an idea of the burden of the disease. If you had 10% of the people already infected in New York, and the deaths are something like 200 or 300, these numbers are close to influenza deaths caused yearly. If the mortality is closer to what an influenza like disease has, we should be concerned of course, but the concern is much weaker than if the death rate is much higher. Here [In the COVID-19 case] the current estimate of mortality rate is between 2% and 3%, which is a factor of 20 to 30 more than influenza, which is 0.1%. This depends on counting how many people are infected in the first place. We don’t have a good sense for it. Testing can help ascertain how problematic this disease is. Testing also allows us to find out who is sick so that we can isolate them.

Would you be interested in working with Indian scientists?

My group has offered help to Indian scientists. We have just initiated discussions with a number of institutions in India. We would be very happy to work with Indian scientists in their quest to respond and build mathematical models and support the assessment of interventions. India has a strong scientific base and this should also allow us to expose the new class of models and computational techniques we have developed in the Indian context. We will have a lot to benefit ourselves too. Not just perspective, but we may be able to carry out interesting studies in India that can help policymakers come up with better regulations and hopefully better response methods.

Related Topics
This article is closed for comments.
Please Email the Editor

Printable version | May 15, 2021 5:31:16 PM |

Next Story