Covid-19: Where is India Heading? (A Data Science Perspective)

Confessions of an Addict

I am a Covid-19 addict.

I find myself compulsively gathering information regarding the Coronavirus outbreak when I should be spending time at work. It gives me the kind of guilty pleasure that I suppose a porn or chocolate addict gets from devouring forbidden items.

What are the reasons for the addiction? First of all, of course, there is the fascination of the macabre — the thrill of watching a disaster movie playing out in real-time. It is Contagion over again except that this time it is happening not on the movie screen but around you.

Then again, the media and your friends feed the addiction. The Covid-19 outbreak seems to be the only topic of discussion for news channels and social media. Several news outlets, which normally hide behind a pay wall, have now allowed Coronavirus related material to be freely accessible, in the public interest, thus providing the drug that feeds the addiction.

But the real reason, I believe, is that the Covid-19 outbreak provides any data scientist worth his salt a treasure trove of data for analysing a public crisis in real time. It may be selfish and perverse — but I can’t help feeling a certain surge of excitement at seeing how good the crisis has been in making abstract mathematical concepts come to life.

I have, for example, never before come across such a compelling example of exponential growth.

Image for post

Source: https://www.worldometers.info/coronavirus/country/italy/(accessed on 22nd March 2020)

The Covid-19 crisis has blown to smithereens the hype about Big Data having killed Sampling Theory.

The shape of a curve has now become a matter of life and death for millions! Thanks to the Coronavirus crisis “flattening the curve” has entered the global lexicon.

Virus Mathematics

The underlying mathematics of the infection is simple conceptually, though complex when one gets down to the details. The clearest explanation that I have so far come across is in the following two articles by Tomas Pueyo.

https://medium.com/@tomaspueyo/coronavirus-act-today-or-people-will-die-f4d3d9cd99ca

https://medium.com/@tomaspueyo/coronavirus-the-hammer-and-the-dance-be9337092b56

Like all disease outbreaks the Covid-19 would run its course and peak at some point even if no precautions at all were taken. This is the situation illustrated in the red graph to the left below.

Image for post

Source: https://www.livescience.com/coronavirus-flatten-the-curve.html (accessed on 22nd March 2020)

With intervention, to identify the infected early & reduce the mingling between the infected and the uninfected, it is possible to “flatten the curve”. The curve to the right is flatter and the peak occurs at a later date. While the total number of infections may remain the same in both cases, in the second scenario there are fewer cases at any point of time so that the healthcare system is not overwhelmed.

Flattening the curve would also buy scientists more time to develop a cure or vaccine and hospitals and governments to stock up on medical and other necessities.

Flattening the curve saves lives. Puyeo’s calculations suggest that Covid-19 death rates for Iran and Italy — where the health care system was overwhelmed by the crisis — are ~3%-5% while that for South Korea — which managed to significantly flatten the curve — is ~0.5%.

Business as Usual

Why then did so many countries go about business as usual, in spite of overwhelming evidence that intervention — especially in the early stages of the epidemic — saves lives?

Interventions can loosely be grouped into two categories:

1. Test, trace and treat: this includes testing suspect cases, contact tracing, quarantine (home stay) of people who might be infected and isolation and treatment of those who are known to be infected.

2. Social distancing: taking steps to reduce possible contacts between the infected and uninfected populations. These could range from relatively small measures such as drawing markers on the floor to space people who stand in supermarket queues to ban on large gatherings to complete lockdown of shops, businesses and transport.

Interventions however come at a cost. There is first of all the direct cost. Just the cost of each Coronavirus test is ₹4,500 (approximately US $ 60) in India, for example. The indirect economic cost of measures such as shutting down travel, entertainment and shopping would of course be far higher.

Image for post

Some shoppers keeping their distance from one another at a FairPrice supermarket in Singapore on March 21, 2020. (Source: ST (Straits Times) PHOTO: JASON QUAH)

Puyeo in his article picturesquely likens interventions to a hammer blow. It suppresses the virus but also hits the economy. The timing and force of the blow must be right.

Singapore by very finely tuning its interventions has managed to keep life running more or less normally while containing Covid-19. Schools, cinemas, supermarkets and offices are all open. However, as Singapore is a highly globalized city, its economy is under severe strain, with the travel and tourism sectors virtually moribund and retail and F&B suffering gravely.

While Singapore has finely calibrated its response to Covid-19, Thailand — a country which depends heavily on the tourist dollar — has waffled and changed course many times. It introduced and relaxed border restrictions several times in a puzzling sequence and began to test widely only when evidence of exponential spread of the disease became abundantly clear. As a result it now faces a serious crisis that, according to a Thai doctor, could result in 7,000 deaths and 350,000 cases by mid April.

https://www.bangkokpost.com/thailand/general/1884640/stay-home-or-share-italys-covid-experience

India - the Next Hotspot?

It was becoming painfully clear by the middle of March that the vast majority of countries had failed to cope effectively with the crisis. The United States fell short. However, so did Canada and Western Europe which had prided themselves on their universal healthcare systems. The Eastern Democracies (Singapore, Taiwan, Japan and South Korea) on the other hand, were doing much better. China too (after having created the mess in the first place by failing to contain the virus in Hubei province) turned things around much sooner than anyone expected.

The reasons for the relative performance (or lack of performance) of different countries will surely continue to be studied for years to come. But given the situation, it was unsurprising that experts of all stripe began predicting that India would be the next Covid-19 hotspot.

https://www.stuff.co.nz/national/health/coronavirus/120478392/coronavirus-experts-fear-india-will-be-the-next-covid19-hotspot

“Thank God this didn’t start somewhere like India” remarked Jim O’Neill the former Goldman Sachs Chief Economist. O’Neill’s comment was tactless but there was genuine cause for concern. India is a relatively underdeveloped and chaotic society and its record in public health care is spotty at best. For example, India reports approximately 220,000 deaths each year due to tuberculosis — about a quarter of the world’s total. Could India possibly cope with an emergency that was overwhelming a large part of the First World?

However, there were also reasons for hope. The Indian record in disaster management is, in some respects, exceptionally good. For example, India has done a phenomenally good job in recent years at limiting death and damage from natural disasters such as cyclones. Thirty years ago a major cyclone would take tens of thousands of lives. Nowadays the numbers are typically in the low double digits. This is certainly no mean achievement given that the Americans lost 1,833 souls to Hurricane Katrina.

India had a Plan

India too had its exponents of the business as usual model and their arguments did have considerable substance. As we have already noted, more than 200,000 people die from TB each year in India. This is a lot more than the total number of deaths from Covid-19 worldwide till date. Wasn’t Covid-19 getting undue importance, they would argue, simply because TB mainly affects the poor.

The prevalence of TB in India is surely a scandal — if not an outright crime. However, having said that, it is important to note that there are important differences between TB and Covid-19. Unlike Covid-19, TB has a long incubation period. Its treatment usually does not require hospitalization. As a result, TB cases do not tend to clump together and overwhelm the healthcare system the way Covid-19 cases might.

India’s reported its first Covid-19 case in the southern state of Kerala on 30th January. Two more — also in Kerala — were reported during the next couple of days. After that there was a lull. India took advantage of this to put a plan in place by 26th February.

http://dgmhup.gov.in/documents/containment.pdf

The plan had two parts. In the first phase — the focus would be on stopping the import of Covid-19 cases from abroad. In the second, to be put in effect when limited local transmission had commenced but the virus was not yet widely circulating in the Indian population, the focus would be on cluster containment. The cluster containment strategy would be to contain the disease within defined geographic area by early detection, breaking the chain of transmission and thus preventing its spread to new areas.

The Storm Breaks

The first part of the plan was put in motion when the next Covid-19 cases were detected in India in early March. The initial cases were all travel related. The Government of India quickly put in place a series of restrictions on international travel which grew more and more draconian over a period of 3 weeks, culminating in virtually complete closing of all borders on 22nd March.

There were quite expectedly early hiccups. Indian governance is a patchwork of central and state rule. Healthcare is a joint responsibility of the Centre and the states. For the last 6 years India has been rules by the BJP while many of the states are ruled by other parties. Yet, a surprising amount of unity was achieved in a short period. Narendra Modi, the Prime Minister of India, Arvind Kejriwal the Chief Minister of Delhi and Mamata Banerjee, the Chief Minister of West Bengal belong to three different parties and rarely see eye-to-eye. Yet they united in sending the same message to the people.

The message was loud and clear. Avoid public gatherings; go out as little as possible; practice social distancing; if you have come back from a foreign trip please quarantine yourself at home for 14 days. Celebrities who broke the rules and continued to perform after their return from trips abroad, were named and shamed on both mainstream and social media. Companies — including ours — put in place arrangements for working from home.

Some of the ruling party functionaries came up with decidedly bizarre methods to confront the crisis. The President of the BJP unit in West Bengal claimed that drinking cow urine would render people immune to Coronavirus. The claim was widely circulated and justly derided. A mass gathering had been planned in Ayodha on the occasion of Ram Navami on 2nd April. A senior BJP functionary had said that the gathering could go ahead since Lord Ram would protect his devotees. Fortunately, saner voices prevailed and the gathering was cancelled.

Image for post

Source:Drawn using data from https://en.wikipedia.org/wiki/2020_coronavirus_pandemic_in_India which in turn acknowledges Worldometer & MoHFW as its data sources. )

The virus continued to spread. By the third week of March it was clear that India, was fighting against exponential growth exactly as other countries had done. The number of cases seemed to be doubling approximately every 5 days. Left unchecked, there would be about 3,000 cases in 2 weeks, about 25,000 cases in a month and 200,000 cases in 6 weeks. In 2 months we would be looking at a mind boggling 1.6 million cases.

Clearly it was time to move on to cluster containment to try to flatten the curve.

On 22nd March India brought down the hammer and brought it down hard. Lockdown was imposed on 75 of the 718 districts of India. Virtually all factories and shops were closed. Gatherings of more than 7 people were banned. Only services deemed essential such as health services, police and the fire brigade continued to operate. Public transport practically ground to a halt.

Test, Test, Test

“We have a simple message to all countries — test, test, test,” so said the WHO Director General Tedros Adhanom Ghebreyesus at a press conference in Geneva on 16th March. The initial Indian Covid-19 response has been much criticised because of the limited testing performed in the first phase.

https://www.aljazeera.com/news/2020/03/india-poor-testing-rate-masked-coronavirus-cases-200318040314568.html

It is hard to tell whether the criticism is warranted or not.

The Director General’s advice is futile for countries which might not have test kits or other resources necessary for testing.

Ideally everybody — and that means all the 7 billion people in the world — should be tested for Covid-19, since even people without symptoms may be infected and help to spread it. This is clearly impossible.

Image for post

Source: https://ourworldindata.org/grapher/covid19-tests-per-million-people?country=IND+TWN+ISR+KOR+United%20States%20-%20CDC%20samples%20tested+VNM+IRN+JPN

As the above chart shows no country comes even close to testing its entire population. The South Koreans lead the pack, having performed more than 300,000 tests, in a country of 51 million people. (In per capita terms a few small countries like Bahrain do better.)

So how do we determine how many to test, in order to establish with a reasonable degree of confidence the proportion infected in the population? And make sure that we have enough resources to do so? This is where Sampling Theory comes into its own.

Statisticians have been studying this kind of problems for a hundred years or more and Sampling Theory is the answer that they have come up with. It has been claimed in recent years that Big Data Analytics has killed Sampling Theory or at least driven it to the margins of data science. The Covid-19 crisis has however exploded these myths and showed that the rumours of death of Sampling Theory — like Mark Twain’s — were highly exaggerated.

I have neither the data nor the expertise to determine whether enough testing was carried out in India in early March. However, I suspect that more testing would not have bought too much extra time before the hammer came down. Perhaps with more extensive testing the presence of local transmission might have been detected a week earlier and lockdowns started that much sooner though this is sheer speculation.

Through a Glass Darkly

I do not possess a crystal ball and so do not know how the war on Covid-19 will go in India.

There is however ground for optimism. India -unlike many other countries- is proceeding according to a definite & clear plan. The usually fractious and argumentative Indians are displaying rare unity of purpose. Perhaps Italy served as a poke in the eye.

The Indian strategy — as far as one can infer from the containment plan and the actions taken so far — is to impose near simultaneous lockdown across country and then halt all domestic travel. This was also tried in Italy but that move came much too late in their game. It may however be hoped that large tracts of India are still unaffected by the virus. If that is the case then, 2–4 weeks of lockdown may suffice to flatten the curve. The disease would be locked in a few discrete clusters, where it can be mopped up like the remnants of a beaten army. It may be necessary to keep some restrictions — especially those on foreign travel — in place somewhat longer but, with just a bit of luck, India may escape the worst.

If the Indian strategy succeeds it may provide the poorer countries of Asia, Africa and Latin America an excellent template for epidemic control.

However, IF is the operative word. India’s success is by no means assured.

As a practitioner who has for more than 20 years been applying mathematical models to solve real- world problems, I realize well that while good modelling is critically important, implementation is far more so.

The challenges are many.

It will be necessary to ensure sufficient medical resources for the emergency. At this point India supposedly has 150,000 test kits. That may not be enough. It will also require ventilators, protective gear, isolation facilities, medicines, doctors, nurses and many things more.

It will also be necessary to ensure that essential services and supplies are maintained throughout the lockdown.

The hammer of lockdown will hit the poor the hardest. Kerala, always ahead of the curve in health matters, has announced a slew of relief measures. However, other Indian states have so far acted in a very piecemeal fashion when it comes to providing relief.

The worst crisis will be faced by migrant workers from the rural hinterland who flood India’s cities and towns looking for work. When work dries up — as it must during a lockdown — they will try to head back their villages, to be close to food, family and community support. This is what they always do in a crisis.

Except — and in this respect this crisis is different from all previous ones — this time they will not be able to go back. They will not — must not — be allowed to return to their villages to spread possible infection and create new clusters. It would therefore be necessary to create temporary shelters to house millions of people and provide them sustenance.

The outcome of the war will be known in about 30 days. But I remain cautiously optimistic.

Disclosure: I am the Director of Smart Consulting Solutions Pte Ltd, incorporated in Singapore and its subsidiary Radix Analytics Pvt Ltd, incorporated in India. I am also a Visiting Faculty Member at the Indian Institute of Management, Udaipur. However, the opinions expressed in this post are solely mine and not necessarily shared by any company or institution with which I am affiliated.

I am grateful to my friend Dr. Ashish Kumar Dawn for sharing the containment plan and many insightful comments.

I would like to draw everybody’s attention to the excellent work by Covid-19 study group of which I only became aware after posting my blog. This is essential reading for anybody who wishes to understand how Covid-19 will affect India. It also prescribes very sensible policies for both the government and the people.

https://medium.com/@covind_19/predictions-and-role-of-interventions-for-covid-19-outbreak-in-india-52903e2544e6

I thank Rohit Chitre for pointing out that kurtosis is unrelated to flattening the curve.

Please share this post if you think your friends will find it interesting or informative. Please also feel free to send your comments to nilotp@yahoo.com or nilotpal@smartcs.sg

I have spent over 30 years in academia and industry exploring how to use mathematical methods to solve real world problems

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store