domeet webmaster

After Plague Company was removed, Stanford produced an epidemic simulator for you

Editor’s note: This article comes from WeChat public account “Xin Zhi Yuan” (ID: AI_era) , edited by Xiao Lin and Meng Jia.

Source: Stanford, etc.

[Guide to Xinzhiyuan] Is social isolation really useful? Trump’s decision to extend the quarantine period to April 30 was a combination of a review of 12 different statistical models. Stanford University biologist Erin Mordecai and a team of researchers have developed an interactive simulator that simulates the spread of COVID-19 over time and vividly demonstrates the role of social isolation in controlling the epidemic.

After Plague Company was taken down, you no longer have any games to simulate the epidemic?

The developers of Plague Inc. have been invited to lecture at the US Centers for Disease Control. It allows players around the world to have a basic knowledge reserve, perceptual knowledge, and crisis awareness of the infection and transmission mechanism of the virus. It is a great popular science game. After the outbreak, it has been removed from official platforms at home and abroad.

Now you have a more accurate outbreak simulator!

Of course, you must have good English to play. Students who have over 100 TOEFL can find links at the end of the article to adjust their parameters. This article only introduces the research conclusions made by scientists using this model. >

Stanford University biologist Erin Mordecai and a team of researchers have developed an interactive website that makes various interventions such as social isolation and quarantine into smoothly adjustable knobs, simulating COVID-19 over time The spread curve allows people to see the dynamic impact of different measures on the number of cases, the number of deaths, and the comparison with the hospital load.

Stanford University biologist Erin Mordecai

Research results: visually show the role of social isolation

The above four figures show the direct reduction of the number of hospitalized patients under the implementation of social isolation measures of 0, mild, moderate, and severe. The red line in the picture is the carrying capacity of the hospital.

1. Reduce curve peaks: Many of our health resources have a fixed capacity. If there are too many cases at the same time, we will not be able to take care of everyone who is sick. Exceeding the upper limit of the number of hospitalizations will mean having to decide which patients to prioritize, while others cannot be treated, and we have seen this happen in Italy and other places. The greater the social distance we practice, the flatter the curve will be.

2. Delay peaks: Over time, many resources will become more and more abundant. We will be able to produce more resources such as treatments, ventilator and hospital beds. Practices such as social alienation can help delay the peak of cases and earn us time to get the resources we need.

3. Preventing outbreaks at the end of the curve: once the number of cases begins to decrease, can we stop social isolation? Do not. These three graphs show the situation of strong social isolation lasting 3/5/10 months. It is clear that if we remove control too quickly, then we may see the spread of disease reappear and the number of cases rebound rapidly because many people Still susceptible. Desegregation after 3 or 5 months can lead to a surge in the number of patients that can even significantly exceed the capacity of the medical system.

We saw this in the 1918 flu pandemic, when many US cities lifted restrictions after 3 to 8 weeks and saw the second largest flu peak.

To prevent the resurgence of COVID-19, we need to take multiple interventions over 12 to 18 months or longer, until effective treatments and / or vaccines are widely used.

Alternative Solution to Social Isolation for Months: Lighting Switching Method

We don’t need to be isolated for a full year or more. Experts have shown that if we use interventions that actively turn on and off, such as The Lightswitch Method, weA relatively low transmission rate can be maintained, allowing a greater range of activity, while still maintaining the epidemic at a level that our healthcare system can manage.

When we turned on the lights, we started to maintain social distance, and the number of cases began to decrease. When we turned off the lights, people could resume socialization to a certain extent and make minor adjustments. The number of cases will start to increase a bit, but not out of control until we regain social distance. Based on the data, we can set the switching time to a certain time (three weeks on, three weeks off) or a certain threshold (for example: open when there are 15 inpatients in a week, and less than 2 inpatients in a week Close in case).

In this way, we can find a balance between preventing the spread of the new crown and normal life.

Dynamic modeling of infectious diseases

The basic principles of statistical models are relatively obvious. Epidemiologists divide the population into different sections. The SIR model is a common mathematical model for describing the spread of infectious diseases. The basic assumption is to divide the population into the following three categories:

1 Susceptible: Refers to those who are not sick, but lack immunity, and are susceptible to infection after contacting the sick.

2 Infective: Refers to a person infected with an infectious disease, who can spread to susceptible people.

3 Removed: The person who was removed from the system. People who become ill (immunized) or die. These people are no longer involved in the infection and infection process. In the SIR model, there are two transformational relationships between the above three types of population.

For people who are “exposed” but not yet infected, some models will also add E to become SEIR models. Modelers then formulate variables based on their judgment of the spread of the disease and then run. These variables include: how many people an infected person has infected or recovered before death, how long it takes for an infected person to infect another, and so on.

SEIR model legend

“Everyone was susceptible at the beginning. The number of infected people was very small. They infected the susceptible people, and the number of infected people began to increase exponentially.” Helen, an infectious disease epidemiologist at Boston University School of Public Health · Jenkins said.

Trump extended the quarantine period to April 30, based on 12 statistical models

On March 29th, at a press conference held by the White House, US President Trump repeated the forecast of the US crown ’s total death data “2.2 million” 16 times. In a previous study published March 25 at the Institute of Health Measurement and Evaluation (IHME) at the University of Washington, it was predicted that 81,114 people in the United States may die from neocoronary pneumonia in the next 4 months. The inflection point of the epidemic was around the second week of April. IHME is a powerful data processing agency with about 500 statisticians, computer scientists, and epidemiologists. Chris Murray, person in charge of IHME: According to the latest model results last week, the overall situation has improved for the first time. The death toll statistics have dropped by 20% compared to last week, and the curve has gradually flattened.

From an epidemiological point of view, the predictions of these death data are derived from mathematical prediction models. White House New Crown Virus Coordinator Dr. Deborah Birx said at a press conference that Trump’s decision to extend the quarantine period to April 30 was a combination of a review of 12 different statistical models.

The 9-cell model used by the Stanford team: S = susceptible E = exposed Ip = presymptomatic but infectious state

Ia = asymptomatic infectionIs = severe Im = mild H = hospitalized R = rehabilitated D = dead

Department of Epidemiology, Harvard UniversityMarc Lipsitch’s team also used the SEIR model, adjusting the data to simulate tightening or relaxation of social isolation measures and possible seasonal changes in Covid-19 infection (similar to flu). He changed R0 during the simulation. The basic infection index R0 refers to the average number of people infected by an infected person without external intervention. In this model, stopping social isolation in the strict sense (without the need for measures such as vaccines or treatments) will rapidly increase the infection rate to about two critical cases per 1,000 people, which is equivalent to 660,000 Americans Coming soon. The team’s model found that even with the most stringent bans in place from April to July, the epidemic would pick up around fall.

Original link:

https://news.stanford.edu/2020/03/30/modeling-social-distancings-impact/

Simulation model link:

https://covid-measures.github.io/

Stanford’s magical source code can be found on Github:

https://github.com/morgankain/COVID_interventions

2.2 million people die of new coronary pneumonia? Is Trump’s worst ending reliable? Https://www.cnbeta.com/articles/tech/961793.htm

Dynamic modeling and parameter identification of new coronavirus in Wuhan based on SIR model (with Python code): https://zhuanlan.zhihu.com/p/104645873

The Mathematics of Predicting the Course of the Coronavirus: https://www.wired.com/story/the-mathematics-of-predicting-the-course-of-the-coronavirus/