Hey Habr! I present to your attention the translation of the article by Jeremy Howard and Rachel Thomas.
From the translator
In Russia, the problem of Covid-19 is not so acute at the moment, but it should be understood that in Italy two weeks ago there was no such critical situation. And it is better to inform the society in advance than to regret later. In Europe, many do not take this problem seriously, and thereby put many other people at risk - as is now evident in Spain (the rapid increase in the number of cases).
Article
We are data scientists, our job is to analyze and interpret data. And the data on covid-19 is cause for concern. The most vulnerable groups in our society, the elderly and the poor, are at the highest risk, but to control the spread and impact of the disease, we must all change our habits. Wash your hands thoroughly and often, avoid crowds, cancel scheduled events, and avoid touching your face. In this post, we'll explain why we care - and why you should, too. Corona in Brief by Ethan Alley (President of Non-Profit who develops technologies to reduce the risk of pandemics) is a great article, summarizing all the key information.
We need a working healthcare system
Just a couple of years ago, one of us (Rachel) was diagnosed with a brain infection that kills about a quarter of the people who get it; a third get a lifelong mental breakdown. Many are left with lifelong visual and hearing damage. Rachel arrived at the hospital parking lot in critical condition, but she was lucky to receive the attention, diagnosis, and treatment she needed. Until recently, Rachel was completely healthy. With a high degree of probability, we can say that quick access to the emergency room saved her life.
Now, let's talk about covid-19 and what might happen to people in a similar situation in the coming weeks and months. The number of people infected with covid-19 is doubling every 3-6 days. At a rate doubling every 3 days, the number of infected people could increase 100-fold in just XNUMX weeks (it's actually not that simple, but let's not get too carried away with the details). infected people need many weeks of hospitalization, and many of them need oxygen. Despite the fact that this is only the beginning of the spread of the virus, there are already regions where there are no empty places in hospitals - and people cannot receive the necessary treatment (not only for coronavirus, but also for other diseases, for example, that life-saving therapy, in that Rachel needed). For example, in Italy, where just a week ago the administration said that the situation was under control, now about 16 million people are locked at home (Update: 6 hours after this post, Italy locked down the entire country), and similar tents are being put up to somehow cope with patient flow:
Medical tent in Italy.
Dr. Antonio Pesenti, head of the regional department responsible for crisis situations in northern Italy, : "We have no other choice but to organize intensive care in the corridors, in operating rooms, in the wards ... One of the best healthcare systems - in Lombardy - is on the verge of collapse."
It's not like the flu
The mortality rate for influenza is estimated at 0.1%. Mark Lipstich, director of the Center for Dynamics of Infectious Diseases at Harvard mortality from coronavirus in 1-2%. found a death rate of 1.6% in February in China, 16 times higher than that of the flu (this estimate may be inaccurate, as deaths rise when the healthcare system fails). Positive assessment: 10 times more people will die from the coronavirus this year than from the flu (and (Elena Grewal, former Director of Data Science at Airbnb, shows that in the worst case, 100 times more people could die). And this is not taking into account the huge impact that has been made on the medical system, as described above. It is understandable that some people try to convince themselves that this situation is nothing new and that the disease is very similar to the flu - because they really do not want to accept an unfamiliar reality.
Our brains are not designed to intuitively understand the exponential growth in the number of sick people. Therefore, we must analyze this situation like scientists, without resorting to intuition.
What will it look like in two weeks? Two month?
On average, each person with the flu infects about 1.3 people. This is called the "R0" of the flu. If R0 is less than 1.0, the infection does not spread and stops. At a higher value, the infection spreads. Coronavirus has an R0 of 2-3 today, outside of China. The difference may seem small, but after 20 "generations" of infected people transmitting the infection, 0 people will become infected with an R1.3 of 146, and 0 million with an R2.5 of 36! (This, of course, is very approximate and many factors are ignored in this calculation, but it is a reasonable illustration of the relative difference between coronavirus and influenza, all other things being equal).
Note that R0 is not a fundamental disease parameter. It depends on the response and may change over time. It is noteworthy that in China, the R0 of the coronavirus has decreased significantly - and is now approaching 1.0! How? - you ask. By applying all the necessary measures on a scale that is difficult to imagine in a country like the United States: completely shutting down metropolitan areas and developing a screening system that allows you to monitor the condition of more than a million people a week.
On social media (including popular profiles like Elon Musk) there is often a misunderstanding of the difference between logistical and exponential growth. Logistic growth refers to the pattern of the epidemic spread of the S form. Exponential growth, of course, cannot go on indefinitely - then there would be more infected people than the entire population of the Earth! So as a result, the rate of infection should always slow down, leading us to an S shape (known as a sigmoid) of growth over time. At the same time, a decrease in growth happens for a reason - this is not magic. Main reasons:
- Massive and effective actions of society.
- A high number of infected, which leads to a low number of potential victims due to a lack of healthy people.
So there is no logic in relying on logistical growth as a way to control the pandemic.
Another reason why it is difficult to intuitively perceive the impact of the coronavirus on your local community is the significant delay between infection and hospitalization - usually around 11 days. It may seem like a short period, but by the time you notice that the hospitals are full, the infection will reach a level where there are 5-10 times more infected.
Note that there are some early indications that the impact on your region may be somewhat dependent on the climate. In the article "It says that the disease has so far spread in temperate climates (unfortunately for us, the temperature in San Francisco, where we live, is just in this interval; the main centers of Europe, including London, also fall there).
"Do not panic. Keep calm" does not help
One of the most common responses to calls to be vigilant on social media is "Don't panic" or "keep calm". This, at least, does not help. No one thought that panic was the best way out of the situation. For some reason, however, "keep calm" is a very popular response in certain circles (but not among epidemiologists, whose job it is to keep track of things like this). Perhaps "keep calm" helps someone justify their own inaction or feel superior to the people they represent in a state of panic.
But "keep calm" can easily lead to a failure to prepare and respond. In China, 10 million people were placed in isolation and two new hospitals were built by the time they were in the state of today's US. Italy has waited too long and just today (Sunday 8 March) they have announced 1492 new infections and 133 deaths despite the lockdown of 16 million people. Based on the best information we can confirm at the moment, just 2-3 weeks ago Italy was in the same situation as the US and England are today (in terms of infection statistics).
Note that almost everything related to the coronavirus is up in the air. We don't know the rate of infection or mortality, we don't know how long it survives on surfaces, we don't know if it survives and how it spreads in hot climates. All we have is our best guesses based on the best information we could get our hands on. And remember that most of this information is in China, in Chinese. Now the best way to understand the Chinese experience is to read the report based on a collaborative study of 25 experts from China, Germany, Japan, Korea, Nigeria, Russia, Singapore, USA and WHO.
When there is some uncertainty - that perhaps there will be no global pandemic and that perhaps everything will just pass without the collapse of the hospital system - this does not mean that the right decision is to do nothing. This would be too speculative and sub-optimal in any case. It also seems unlikely that countries like Italy and China would shut down huge parts of their economies without a good reason. And this is not the same as what we see in infected areas where the medical system fails (for example, in Italy, 462 tents are used for pre-examination, and patients from intensive care were from infected areas).
Instead, the thoughtful, sensible answer is to follow the steps recommended by experts to prevent the spread of infection:
- Avoid crowds.
- Cancel events.
- Work remotely (if possible).
- Wash your hands when you get in and out of the house—and often outside the house.
- Don't touch your face, especially outside the home (not easy!).
- Disinfect surfaces and bags (it is likely that the virus survives up to 9 days on surfaces, although this is not known with certainty).
This applies not only to you
If you are under 50 and do not have risk factors such as a weak immune system, cardiovascular disease, smoking or other chronic illnesses, then you can relax: it is unlikely that you will die from the coronavirus. But how you react is still very important. The chance that you will get infected is still high - and if you get infected, the chance of infecting others is also great. On average, each infected person infects more than two people, and they become contagious before symptoms appear. If you have parents you care about or grandparents and you plan to spend time with them, you may later turn out to have infected them with the coronavirus. And this is a heavy burden that will remain for life.
Even if you don't have contact with people over 50, chances are you have more colleagues and acquaintances with chronic diseases than you realize. that few people talk about their health at work because of . We are both at risk, but many of the people we interact with may not know this.
And, of course, this applies not only to people in your environment. This is also a very significant ethical issue. Everyone who makes an effort to slow the spread of the virus is helping the entire society in reducing its spread. As Zeynep Tufekci wrote : "Preparing for the near-imminent global spread of the virus... is one of the most socially beneficial, altruistic things you can do." She continues:
We must prepare, not because we personally feel in danger, but also to reduce the danger to each of us. We must prepare not because the end of the world is coming, but because we can change every aspect of the risk we face as a society. It's true, you need to get ready because your neighbors need it - especially your elderly neighbors, your neighbors who work in hospitals, your neighbors with chronic illnesses, and your neighbors who can't prepare themselves due to lack of time or resources.
It affected us personally. The biggest and most important course we've done at fast.ai, which represents the culmination of years of our work, was due to start at the University of San Francisco in a week. Last Wednesday (March 4) we decided to take the entire course online. We were one of the first courses to switch to . Why did we do it? Because we realized early last week that by doing this course, we are indirectly encouraging the mass gathering of hundreds of people in an enclosed space, many times over several weeks. Gathering groups of people in an enclosed space is the worst thing to do in this situation. We felt obligated to prevent this. This decision was extremely difficult. The time spent working with students has been one of the greatest joys and the most productive period of each year. And our students were going to fly in from all over the world for this course - we didn't want to disappoint them.
But we knew that this decision was the right one, because otherwise we would most likely increase the spread of the disease in our society.
We must flatten the curve
This is extremely important because if we reduce the spread of infection in a society, then we will give hospitals in that society time to deal with both the infected patients and the regular patients they are supposed to treat. This is called "flattening the curve" and is clearly shown in this diagram:
Farzad Mostashari, former National Health IT Coordinator, explained: “There are new cases every day with no travel history or connections to known cases, and we know they are just the tip of the iceberg due to delays in screening. This means that in the next two weeks the number of infected will increase significantly ... Trying to impose small restrictions in the face of exponential spread is like concentrating on sparks when the house is on fire. When this happens, you need to change the strategy to mitigating precautionary measures to slow the spread and reduce the impact on public health.” If we can bring the spread down enough that our hospitals can handle the strain, then people will have access to treatment. But if there are too many sick people, many of those who need hospitalization will not get it.
Here's what it looks like in terms of mathematics Liz Specht:
In the US, there are 1000 hospital beds per 2.8 people. With a population of 330 million, we get about a million seats. Usually 65% of these places are occupied. This leaves us with 330 thousand free hospital beds throughout the country (perhaps a little less during this period, taking into account seasonal diseases). Let's take the numbers in Italy as a basis and assume that 10% of cases require hospitalization. (Keep in mind that for many patients, hospital stays last weeks—in other words, turnover will be extremely slow as beds fill with coronavirus patients.) According to this estimate, by May 8, all empty places in US hospitals will be filled. (Naturally, this doesn't say how well hospital beds are equipped to isolate patients with a highly contagious virus.) If we're wrong about the number of serious cases, this only changes the filling time for hospital beds, by 6 days in each direction. If hospitalization is required in 20% of cases, places will run out ~May 2. If only in 5% - ~ May 14th. 2.5% brings us to May 20th. This, of course, assumes that there is no urgent need for hospital beds (not for coronavirus), which is questionable. The health care system is overburdened, there is a shortage of prescriptions, etc., people with chronic diseases, who are usually self-sufficient and self-organized, can become seriously ill requiring intensive medical care and hospitalization.
The difference is in the reaction of society
As we have already discussed, this math is not exact - China has already shown that it is possible to reduce the spread with emergency measures. Another good example of a successful response is Vietnam, where, among other things, a national advertisement (with a catchy song!) quickly mobilized society and convinced people to change their behavior to more acceptable in this situation.
This is not just a hypothetical situation, as was clearly seen during the 1918 Spanish flu. In the US, two cities have shown very different responses to the pandemic: Philadelphia held a planned parade of 200.000 people to raise funds for the war; San Luis activated a strategy to minimize social contact to reduce the spread of the virus; all public events were cancelled. And this is how the statistics on the dead looked in each of the cities, as shown in :
Different reactions to the 1918 Spanish flu
The situation in Philadelphia quickly got out of control to the point where to bury so many dead.
Richard Besser, former director of the Centers for Disease Prevention and Control during the H1N1 pandemic in 2009, that in the United States "the risk of danger and the ability to protect yourself and your family depend on income, access to medical services, immigration status and other parameters." He points out that:
The elderly and people with disabilities are at increased risk when their daily rhythm and support systems are not working. Those without access to health care, including villages and local communities, will also be affected by distance to nearby centers. People living in closed areas - in social housing, prisons, shelters, or even the homeless - can get infected in waves, as we've seen in Washington state. And the vulnerabilities of the underpaid with undocumented workers and unstable schedules will be exposed during this crisis. Ask the 60 percent of the U.S. hourly workforce how easy it is for them to take time off.
The American Bureau of Job Statistics shows that people in the lower pay tier have paid sick leave.
Most low-income Americans don't have sick pay, so they have to go to work.
We do not have reliable information on Covid-19 in the US
One of the biggest problems in the US is the lack of inspections; and the results of the audits carried out are not properly published, which means that we do not know what is really going on. Scott Gottlieb, the previous head of the Food and Drug Administration, explained that Seattle had better inspections, so we have information about the infection in that zone: “The reason we learned early about Covid-19 infections in Seattle are the focus of independent researchers. There has never been such a complete surveillance in other cities. So other hotspots in the US may not be found at the moment.” According to , Vice President Mike Pence promised that about 1.5 million tests would be available this week, but in the entire US, only 2000 people have been tested to date. Based on work from , Robinson Meyer and Alexis Madrigal of The Atlantic say:
The information we have gathered suggests that America's response to covid-19 and the infection it causes has been shockingly slow, especially compared to other developed countries. The Centers for Disease Control and Prevention confirmed 8 days ago that the virus was spreading within the American community - that it was infecting Americans who did not travel abroad themselves and did not come into contact with anyone who did. In South Korea, more than 66.650 people were tested in the first week after the first internal infection - and soon learned to test 10.000 people a day.
Part of the problem is that it has gone political. In particular, Donald Trump has made it clear that he wants to keep the "numbers" (that is, the number of infections in the US) low. (If you want to learn more about this topic, read the article on Data Science Ethics ""). Head of Artificial Intelligence at Google, wrote about the problem of political disinformation:
When I worked at the WHO, I was part of the international AIDS program - now UNAIDS - created to fight the AIDS pandemic. Employees, doctors and scientists, were completely focused on solving this problem. In times of crisis, clear and precise information is needed to help everyone make informed decisions about how to act (country, state, local government, companies, non-profits, schools, families and individuals). With the right information and measures to listen to the best experts and scientists, we can overcome challenges like HIV/AIDS or COVID-19. With disinformation coming from political interests, there is a real threat of making things worse because we do not respond quickly and decisively during a growing pandemic and because we actively support behaviors that contribute to the spread of the disease by far faster. It is unbearably painful to watch this situation unfold.
It doesn't look like politicians are looking to change things when it comes to transparency. Secretary of Health Alex Azar, “I started talking about the tests that medical workers do to see if a patient is infected with the new coronavirus. The lack of these tests meant a dangerous gap in epidemiological information regarding the spread and severity of the disease in the US, exacerbated by a lack of government transparency. Azar was trying to say that new tests have already been ordered and only quality control is missing to get them. But, they continue:
Trump then abruptly interrupted Azar. “But I think, and this is important, that any person who needed a test today or yesterday got this test. They are here, they have tests and the tests are great. Anyone who needs a background check gets a background check,” Trump said. It is not true. Vice President Mike Pence told reporters that in the United States, demand for tests exceeds supply.
Other countries are reacting much faster and more significantly than the US. Many countries in Southeast Asia are performing well, including Taiwan, where R0 reached 0.3, and Singapore, which . But now it's not just Asia; in France, for example, any gathering of more than 1000 people is prohibited, and schools are closed in three zones.
Conclusion
Covid-19 is an important social issue and we can – and should – work to reduce the spread of the disease. It means:
- Avoid large crowds
- Cancel events
- Work from home whenever possible
- Wash your hands when you get in and out of the house—and often outside the house.
- Avoid touching your face, especially outside the home
Note: Since it was imperative to publish this article as soon as possible, we were not as careful about compiling the list of citations and papers on which we based.
Please let us know if we have missed anything.
Thanks to Sylvain Gugger and Alexis Gallagher for the feedback and comments.
Source: habr.com