According to data released by the National Health and Reform Commission, the number of confirmed cases of pneumonia in the country’s new coronary virus infection increased by 3,235 on The same day on February 3 (0-24). Of these, 2,345 cases were confirmed in Hubei Province. As of 9 A.M. on 4 February, a total of 20,472 confirmed cases, including 2,788 cases of severe illness, 425 cumulative deaths, 633 cases of cured discharge, and 23,214 suspected cases were reported nationwide.
The pneumonia outbreak of the new coronavirus has reached far more than the SARS outbreak of 2003, which has aroused the attention of many countries worldwide.
Figures . . . Changes in the number of confirmed and suspected cases in the last 5 days (unit: example, information from the official website of the National Health and Health Council)
Figures . . . Changes in the number of new confirmed cases and new suspected cases in the last 5 days (units: examples, information from the official website of the National Health and Health Council)
Changes in data can be easily assumed that the growth of suspected cases is controlled accordingly and that the overall upward trend is declining. However, the upward trend of confirmed case data still does not reflect the state of “head down”, and even in the last two days there is a “head up” trend. For this reason, many scholars and research teams are considering the best and worst cases based on past epidemiological studies and what scientists know.
Ma Zhaoyuan’s team at Tsinghua University’s Future Lab predicts that the number of infections will peak between February 12-19, with an outbreak turning point around February 16, with a cumulative number of confirmed at a maximum of 42,000 to 60,000, before gradually decreasing and eventually subsiding. Dr. Ma Zhaoyuan, lead researcher at the Future Lab, told DeepTech that his team had built a dynamic model based on the data released by the country, taking into account existing isolation, testing and treatment methods (the most optimistic estimates and general estimates), as well as the impact of the return passenger flow during the Spring Festival. At the same time, the Tsinghua Medical College’s vaccine team has provided recommendations.
Figures . . . Based on projections of the outbreak data as of 24 hours on February 1, Ma said: “The error in the forecast on February 2nd is five out of a thousand. (Source: Ma Zhaoyuan)
The best estimate is that China’s anti-death control measures are in effect, so the number of infections will be reduced, and the worst estimate is that the outbreak could reach up to 190,000 people in all sorts of bad situations, the journal Nature reported on January 31.
Behind these scenarios is a mathematical model giving predictive data. It must be emphasized that there is a common sense in the field of statistics: all models are wrong. Any model is the result of certain assumptions and reasoning, so there is a certain confidence interval. The complexity and uncertainty surrounding this outbreak makes it almost impossible to accurately predict the development of the outbreak.
But mathematical modeling is meaningful, good, relatively close to the actual model through continuous iteration, can give a gradual correction of the prediction and prevention and control reference. Experts give the best and worst future projections based on previous models and weigh ingest the current situation.
How many people will the new coronavirus infect?
China has long blocked Wuhan, the central city where the pneumonia outbreak has spread, and Chinese researchers quickly shared data on the virus with researchers from the World Health Organization and other countries. But the number of confirmed cases has been rising, to more than 20,000. Based on these updates, a model at Northeastern University’s Biological and Social Technology Systems Modeling Laboratory predicts that the virus could infect 39,000 people out of a population of nearly 30 million in and around Wuhan.
Ian Mackay, a virologist at the University of Queensland in Australia, believes the virus is spreading too far and too fast to be effectively contained.
Ben Cowling, an epidemiologist at the University of Hong Kong, said: ‘In the best case, the number of people infected will be reduced as the effects of the control measures are starting to be reflected.’ However, it is too early to say whether efforts to segregate people and whether the widespread use of masks will work. He said the virus’s incubation period (up to 14 days) is longer than most control measures have been implemented, so it remains to be seen.
Jonathan M. Read of the Centre for Health Information, Computing and Statistics at Lancaster University in the UNITED Kingdom, together with researchers at the Institute of Biological and New Pathogens at the University of Florida and the University of Glasgow’s Virus Research Centre at the British Medical Research Council, worked together to develop another predictive model: in the worst case scenario. , Wuhan is likely to infect up to 190,000 people. In addition, scientists are particularly concerned about new outbreaks outside China. The virus has now spread to southeast Asian and European countries, as well as small local populations in the United States, and local authorities have quickly isolated those affected.
Figures . . . WHO said on January 31st that the pneumonia was “human-to-human” outside China (Source: Twitter)
Who Director-General Tan Desai, in declaring the pneumonia outbreak as an “international public health emergency of concern ” (PHEIC)” press conference site, repeatedly stressed that this announcement is not because of the outbreak in China, mainly to enable other countries to better prevent and control the outbreak, Especially to protect countries where some health-care systems are fragile. WHO has great confidence in China’s ability to contain the spread of the epidemic, and the scale and commitment of the Chinese government to this level of epidemic response is unprecedented and well-deserved.
On the same day as Nature reported, a new paper published in The Lancet speculated that as of January 25, 2020, the new coronavirus in Wuhan may have infected 75,815 people, the basic number of infections of the new coronavirus R0 is 2.68, the infection multiplier time is 6.4 days. The estimated imported cases from Wuhan in Chongqing, Beijing, Shanghai, Guangzhou and Shenzhen were 461 (227-805), 113 (57-193), 98 (49-168), 111 (56-191) and 80 (40-139), with a confidence range of 95%. And human-to-human cases already exist in several major cities.
Figure: As of January 25, 2020, the number of cases that Wuhan exported to some cities in China, as estimated by the model, FOI is infectious (Source: The Lancet)
The researchers used mathematical models based on 2019-nCoV case data reported by the China Center for Disease Control and Prevention (December 31, 2019 to January 28, 2020), and added domestic and international travel data (including trains, planes and highways) and Tencent databases in more than 300 cities in mainland China Demographic data to estimate the scale of the spread of the epidemic, taking into account the impact of the severe quarantine in Wuhan and surrounding cities from 23 to 24 January 2020.
Figure: Prediction of the popular curve in different scenarios in Wuhan and five other major Chinese cities in different scenarios of reduced mobility between transportation and cities (Source: The Lancet)
The study assumes that the transmission rate of the new coronavirus will be reduced by 0%, 25% and 50% respectively after the first-level emergency response in Wuhan City and cities on January 23rd (previous studies have shown that non-drug interventions can reduce virus transmission by up to 50%), and this is based on the calculation that if the infection rate is not reduced, the Wuhan outbreak will be reduced by 2020. Peaks will peak around April, with localized outbreaks in mainland Chinese cities lagging by 1-2 weeks.
If the rate of infection in all cities across the country is reduced by 25%, the growth and scale of the local outbreak will be greatly reduced, the peak of the outbreak will be delayed by about 1 month, the scale will be reduced by about 50%, and if the infection rate is reduced by 50%, the R0 value will fall back to about 1.3, in which case the outbreak will slowly grow in the first half of 2020 However, there will be no spikes.
However, given that Beijing, Shanghai, Guangzhou and Shenzhen together account for more than 50% of all outbound international air travel in mainland China, other countries may still face the risk of spreading growth in 2019-nCoV in the first half of 2020. But as WHO says, as domestic controls on the outbreak have increased, the assumptions of all models have been affected by response softened on this scale.
Michael Levitt, a 2013 Nobel Prize-winning chemist and professor of structural biology at Stanford University in the United States, analyzed the pneumonia outbreak on February 2 and said “the outbreak will soon be over.” Since Michael’s data analysis was first published by Huifu Technologies, a Beijing-based smart rehabilitation app company, DeepTech confirmed and received a positive response to the professor himself yesterday.
Michael Lev itt’s data comes from the Kaggle website and is checked with data from Thesanddata and Jobtube. He divided the data into Hubei and other areas (non-Hubei Province) and said that the vast majority of deaths were concentrated in the 90 km x 35 km range near Wuhan and Tianmen, Hubei Province, and that while the cumulative number of confirmed cases continued to increase, almost all deaths were in Hubei (over 95%), as shown in the figure below.
Figures . . . Cities with mortality rates greater than 1% in Hubei Province are marked with red circles, and cities less than 0.5% are marked with green circles, and deaths are concentrated within 90 km x 35 km of Wuhan and Tianmen (Source: Michael Levitt)
Michael assumes that the first confirmed case of neo-coronavirus pneumonia occurred on November 29, 2019, the number of confirmed/deaths varying in relation to the number of days, and charts the data from January 22, 2020 to February 2, 2020. In Figure E (see figure below), he adds a linear trend line to the data starting January 29, 2020. The fit is very good for the total multiple, the Hubei multiple, and the other region multiples (correlation coefficient or sqrt (R2) and 0.99). This indicates that the multiple may decrease to 1.0 within a week, after which the number of confirmed and deaths will increase slowly.
Specifically, the rate of deaths on the day is decreasing, except for the previous day. Total deaths and deaths in Hubei province have been on a monotony downward trend since 25 January 2020, while the decline since 29 January 2020 has been linear. Linear extrapolation suggests that the number of new deaths from next week will drop rapidly.
Figures . . . Assume that November 29, 2019 is the data curve for the first case of pneumonia (January 22, 2020 to February 2, 2020): A. Cumulative confirmed cases; Similar deaths; Death rate due to illness; A partial change in the number of cases; Part of the death toll changes, all three lines are: total/Hubei/other areas. (Source: Michael Levitt)
But some industry insiders told DeepTech that Michael Levitt’s data analysis was ideal and that it was hard to say what would happen to the epidemic clinically. This assumption is relatively simple, but based on his data curve in Figure B, it can be inferred that the mortality rate of the outbreak is relatively manageable if it is detected and taken.
Is the mathematical prediction model reliable?
So why are the mathematical models given by the various research teams so different? Is the mathematical model’s prediction reliable?
Let’s start with a specific model case. On January 17th the team at Imperial College of Technology estimated that the number of new coronavirus estrains in Wuhan by January 12 had reached 1,723, based on the number of people leaving the country and the number of people sickfrom abroad.
The logic of this study is very simple, assuming that every Wuhan citizen randomly travel internationally, then the proportion of people from Wuhan who go abroad is the same as the proportion of Wuhan citizens who are ill, and then through the number of people diagnosed abroad, the number of people who were ill in Wuhan at that time was reversed.
In response to this study, Dr. Jiang Qi, Ph.D. in Genomic Medicine and Statistics at Oxford University, wrote a rebuttal, saying that it had little reference for the prediction of the outbreak in Wuhan.
In Jiang’s view, this model ignores some important factors. For example, not all Wuhan residents will go abroad, holding a passport is a basic condition for going abroad, you know, Chinese citizens in 2019 passport holding rate of only 13%, only consider ingress overseas cases for the entire situation in Wuhan will be a large deviation.
In addition, Jiang Said, the origin of the outbreak is Wuhan South China Seafood Market, located in the center of Hankou, so the probability of international travel by nearby susceptible people may be significantly higher than the 19 million people used in their model Wuhan metropolitan area. At the same time, according to the outbreak traceability study, most of the initial cases were related to game consumption in the South China seafood market, which is considered a luxury consumption, with higher probability of high income sprees and greater likelihood of international travel among the people involved. Therefore, Jiang believes that the model’s hypothesis may not hold true.
Current data models are highly dependent on assumptions, and if they are unlikely to hold up, they can cause large differences in the forecast results, and may even have more than 10 times the error. “The actual guiding significance of the outbreak is not great with the results of such a ten-fold deviation. “
Figures . . . Map of the Ebola virus outbreak in West Africa (Source: Wiki)
Take another example of the Ebola outbreak. At the end of 2013, the worst Ebola outbreak in Human History occurred in West Africa. Soon, researchers at many medical institutions around the world began using simple mathematical models to predict the spread of Ebola and to make pessimistic predictions about the spread of the epidemic, even predicting that millions of people would die as a result. In September 2014, for example, the U.S. Centers for Disease Control and Prevention released a study that found that by January 2015, there would be 1.4 million Ebola cases in Liberia and Sierra Leone.
You know, the CDC is the absolute authority on global public health, and this huge forecast sounds amazing at the time.
By 17 December 2014, however, the World Health Organization released data showing that the number of infections (including suspected cases) in three West African countries, including Liberia, Sierra Leone and Guinea, where the Ebola haemorrhagic fever outbreak was raging, was 19,031, with 7,373 deaths. Since then, the number of infections has continued to decline, and on 14 January 2016, who declared the Ebola outbreak in West Africa over.
In the end, the death toll from this outbreak was about 11,000. Looking back at the original forecast figures, you might think that these studies are not reliable. The root cause of the model’s indesperation is that many medical institutions rely solely on existing disease prediction models to estimate deaths in Guinea, Sierra Leone and Liberia, without taking into account the complexity and uncertainty of the spread of the virus.
In fact, any model is the result of certain assumptions and inferences, so there is a certain confidence interval. In fact, one of the things statisticians often emphasize is that all models are wrong. This is for the specific guess results of the model, the guess results and the actual situation is difficult to fully match, so without thinking about the model, or even no model.
But a good model is used. Especially in the current race between people and new coronaviruses, the establishment of a realistic mathematical model for short-term prediction of the epidemic has a certain degree of accuracy, the results can effectively block the spread of the epidemic can be inspired, but also for prevention and control measures, such as how many wards are needed, how many medical personnel and how to distribute medical supplies and so on.
Xiao Yanni, vice dean of the School of Mathematics and Statistics at Xi’an Jiaotong University and director of the Center for The Study of the School of Mathematics and Life Sciences, stressed that any prediction based on mathematical models should not be free from the constraints of conditions, which are the basis of which the public can understand the true meaning behind some seemingly “horrible” predictions.
Therefore, when publishing their findings, researchers must make it clear under what conditions their predictions are obtained. This is the best way to avoid causing public panic.
How many deaths will the outbreak bring?
In the course of an outbreak, it is difficult to predict the number of cases except for the number of cases, and it is also difficult to calculate the fatality rate of the virus (i.e. the proportion of people who die) because records of new cases and deaths are constantly being updated.
To date, 425 deaths out of more than 20,000 infections indicate that the mortality rate for the new coronavirus is just over 2%, which is significantly lower than SARS, which has a fatality rate of 14%-15%. Mark Harris, a virologist at the University of Leeds in the UK, said the known mortality rate of the new coronavirus could be further reduced as mild and asymptomatic cases were identified.
Currently, there are no effective antiviral drugs. Although some international research teams are working on drugs and vaccines, they will not be useful in the short term.
Changes in the number of deaths from illness esfactors will also be affected by how China’s health system responds to the large number of cases, according to Nature. Dots and use a ventilator to ensure that the immune system in the patient’s body receives adequate rehydration and oxygen when it fights the virus.
Sanjaya Senanayake, an infectious disease expert at the Australian National University, says China has built two new hospitals in Wuhan to accommodate infected patients, but if the virus spreads to other parts of the world where resources are scarce, such as low-income areas in Africa, Then the local health system could be in trouble.
If the virus spreads around the world, even if the death rate is not high, the death toll will be significant. Adam Kamradt-Scott, a global health care expert at the University of Sydney in Australia, says the current 2 to 3 per cent mortality rate is still high for infectious diseases. The 1918 flu outbreak (often referred to as the Spanish flu) infected about a billion people, a third of the world’s population at the time, and caused more than 2.5 percent of the deaths of those infected — an estimated 50 million people.
But Scott also said the new coronavirus may not trigger this “apocalyptic” condition because it usually does nuns and healthy people.
On the inflection point of the epidemic, the prediction and opinion of Chinese experts
While mathematicians and statisticians give predictions, China’s disease control and infectious disease experts have also made predictions about the trend of the epidemic.
On January 28th Zhong Nanshan, a member of the Chinese Academy of Engineering and director of the National Center for Clinical Medicine for Respiratory Diseases, ruled that it was difficult to estimate absolutely when the pneumonia outbreak of the new coronavirus infection peaked, but should reach a relatively high level in a week or ten days. Based on this calculation, the outbreak may peak 10 days after the 28th, that is, after the Lantern Festival.
On the same day, Gao Fu, a member of the Chinese Academy of Sciences and director of the Chinese Center for Disease Control and Prevention, also explicitly speculated that the inflection point would occur before and after the Lantern Festival. He pointed out that the occurrence of the virus has its own law, the current prevention and control measures are playing a role, suspected cases are decreasing. “It is expected that the situation may improve on the Lantern Festival. Personally, I’m optimistic, even older than this (the Lantern Festival). “
For infectious disease outbreaks, the inflection point is marked by a decline in both the number of suspected infections and the number of cases.
In this regard, there are experts conservative attitude. Yang Gonghuan, former deputy director of the China Center for Disease Control and Prevention, public health and epidemiology experts in an interview with the interface news, said that looking forward to the epidemic turning point now some too hasty, to follow the objective law, before the inflection point arrived, the number of cases of infection appeared some decrease, are normal phenomenon, a little fluctuation can not explain the problem.
Yang Gong-hwan believes that at the national level, the outbreak has to go through two incubation periods, after the first 14 days of incubation period, the inflection point should come in mid-February or after, followed by the second incubation period, when the daily number of confirmed people will gradually slow down, but the number of confirmed cases will continue to slowly increase.
Feng Zijian, deputy director of the China Center for Disease Control and Prevention, told the Beijing News on January 31st that it was difficult for anyone to make such a judgment. “We are simply watching the implementation of various preventive and control measures calmly and carefully. “
“From now on, the inflection point may soon come, ” Chen Wei, a member of the Chinese Academy of Engineering and a researcher at the Institute of Military Medicine of the Academy of Military Sciences, who has led the team for several days, told the China Science Daily on January 31. But after the first inflection point, will the epidemic have a second peak, the third peak? We still have to prepare for the worst, come up with the best plan, ready for the longest struggle. “