invisible

Going back to facts, news from Russia. So instead of of implementing quarantine the Tzar decided to declare paid holiday for a week in the country and employers must bore the cost. Plus (and this is completely something new across the world in terms of response) bank account having more than 1m RUB in deposits will be taxed.

RatherBeOnATrain

Graph of the rates of Reported Deaths by US states:




(Graph posted on Twitter by former FDA Commissioner Scott Gottlieb, @ScottGottliebMD )

PanAmWT

What you guys described are the empirical models that try to simulate the nonlinear aspect of the nature, which is described by the fundamental prediction differential equation. Namely, when the amplitude is no longer small, use of a constant doubling time would be problematic. This is because non-linear interaction occurs, and when outside forcing such as change of behavior or mitigation efforts yield results. This would be reflected in an increase in doubling time (you could also have decrease if something goes wrong, such as a collapse of heathcare infrastructure, loss of health care workers, or a hurricane forces evacuation of population to a stadium etc).

Time change of any and all physical quantities is governed by the differential equation. That is the fundamental mathematical principal of any natural changes and is the basis of Calculus. All statistical models (what every fancy names they may have) are all empirical models trying to use associations, correlations, or probabilities etc to approximate the solution of the differential equation.. They contain no basic mechanism describing the nature. . They can be very useful as they are data-driven. Some AI system such as artificial neuron network is similar.

In climate studies because we know the basic physics principals we can compute nonlinear terms from physical laws. (In reality there are other difficulties so the climate models predicting global warming also has its own set of problems.) I am not an epidemiologist so I do not know the details how they deal the non-linear effect of a variable doubling time. I assume they have to know R0, which may also vary with time, and use some historical data to tune an empirical model that include the effect of counter measures. .

It is obvious that how to determine the doubling rime is the most critical issue here. The results are very sensitive, because that is the nature of this problem. I assume they use an empirical model (probability or whatever you call it) used to estimate the doubling time. This kind of models are call dynamic-statistical model for us atmospheric scientists (such as climate change models predicting global warming). But if we use only statistical model to predict how much global warming will occur in the next century, we will have much less confidence. So in the COVID-19 case if you apply an empirical model directly to predict the final outcome rather than the doubling time and feed it into the differential equation, you introduce more errors.

(Edit add) Thank you for the opportunity of this discussion. I think this shows that how important the counter measures such as social distancing.are. A small improvement of the doubling time can have huge benefits because the sensitivity nature of the linear solutions.

bobbytables

I also hope Sweden and Netherlands keep their current approaches. There’s extremely limited evidence for doing anything else, and thanks to EURO MOMO we can track the all-cause death rates in these individual countries on a weekly basis, so it’s a good opportunity to collect some better evidence one way or the other.

username

Trevor Noah interview with Dr. Anthony Fauci:


and Dr. Vivek Murthy (former US Surgeon General):

wco81

Asymptomatic a few days earlier to being on ventilator with damaged lungs.

https://www.cnn.com/2020/03/26/healt...deo/index.html

doctoravios

I think there are huge complexities in clinical presentation we are not capturing. At the very least, we should be comprehensively recording symptoms of new onset illness in the community even in the absence of testing. We are not even doing this in the UK.

I have a theory (and I should stress it is just a theory and the only evidence I have for it is my clinical experience and observation so far) that this virus can lead to a wide variety of clinical presentations which can broadly be categorised into 3 groups:

(i) Asymptomatic/minimally symptomatic: no perceived subjective symptoms or very mild symptoms not perceived to be abnormal (e.g. very mild sore throat/ticky feeling in throat/nasal congestion/feeling slightly tired or not quite right).

(ii) Mild/moderate upper respiratory tract: sore throat, coryzal symptoms, possible fever/cough, possible loss of sense of smell/taste. Resolved within a few days or a week or two.

(iii) Moderate/severe bronchopneumonia which may lead to ARDS: this is the group we currently have the most data on. Generally older/underlying medical conditions but not exclusively. For those who do recover, recovery time is of the order of weeks/months.

If this is true then we need to learn about the epidemiology of each type of clinical presentation and the links between them. It is possible that R0/duration of infectivity/illness duration/degree of immunity following recovery varies considerably depending on the nature of clinical presentation and the epidemiological models are failing to capture this complexity by taking the standard susceptible/infectious/recovered approach.

There is also the possibility of new strains of the virus emerging which have varying pathophysiology. This could complicate matters even further.

We urgently need seroprevalence studies. Even if the assays are not 100% accurate (and I suspect to begin with they will unfortunately have a high degree of false negatives and false positives), at population level combined with careful documentation of clinical symptoms and progression in well curated case registers, we could begin to tease apart some of the complexity. We need to go back to basics and start collecting detailed clinical histories from anyone who feels unwell in the coming months.

The main thing I am learning right now is that for some reason COVID-19 does not seem to be behaving like any other previous pandemics we have faced in recent times. There is something unusual about this virus, its epidemiology and its pathophysiology which will take a long time to fully understand.

LapLap

For myself (and my kid) it was as if we both had a standard (snotty) cold before symptoms consistent with COVID-19 set in later. Am still feeling guilty for exposing people during that initial time. I took on board all the advice given (in English and Spanish), complied with the British Government’s and NHS advice (in fact was more cautious than told to be), but by the time I had a temperature and cough - and then later the sensation of chest/lung restriction - which made it clear I was not to leave the house at all, I may well have passed it on.
I kept my distance right from the beginning, but the pressure to keep my child at school was quite intense at the time (this was in the week commencing 9th March). It took a lot of conviction and some bluddy mindedness to keep her from returning on the 11th March. She never did go back to school. (It’s been homeschooling this week along with most of the rest of her class - we’re lucky she’s the right age for this (10), my sympathy for those with younger kids, particularly those who struggle with reading at the best of times)

Each week another symptom considered typical of Covid-19 seems to make it onto the “how do I know if I have it?” lists in the Press and Social Media.

invisible

Is there a website with data visualisation and modelling options so one can plot graphic of how many cases will be if number of cases increased x percent per day, or having an option to model suppressing new cases? Also to forecast if country A has x cases and B - y cases (X>Y) then how many days will it take for B to get to level of A?

I understand that this can be plotted in excel, but if there is a ready option it would be better.,

yosithezet

There may be something here: https://www.tableau.com/covid-19-cor...data-resources

doctoravios

Boris Johnson (UK Prime Minister) has developed symptoms of fever and cough and tested positive for COVID-19:


We are all susceptible. We must all work together to tackle this pandemic.

KathinJax

Kinsa Makes smart thermometers. They have been taking the data generated by the thermometer and manual inputs into their app to generate the rates of feverish illnesses. They have made their map public: https://healthweather.us

For us in Jacksonville, I can see that the rate of feverish illnesses have declined since the peak on March 18th. Looking at the map this weekend was very scary to see Florida lit up red for atypical illness.

Andy Slavitt on Twitter seems to have access to more of their data and has a chart showing the difference between Santa Clara and Miami:

Twitter

Of course he suggests that Miami will be hard hit in the next few weeks.

lupine

https://www.latimes.com/world-nation...kills-hundreds

Hundreds dead and more injured in Iran after ingesting methanol in the (mistaken) belief that it would fight COVID-19. Just a tragedy on top of the other tragedies facing Iran.

invisible

another report from Imperial College

https://www.imperial.ac.uk/mrc-globa...n-coronavirus/

Smiley90

https://globalnews.ca/news/6741037/c...ubsidy-canada/

"Trudeau promises 75% wage subsidy for businesses hit by coronavirus pandemic

The federal wage subsidy for small and medium-sized businesses hit by the new coronavirus pandemic is jumping up to 75 per cent, the prime minister announced on Friday."