Simulating an epidemic

If we can test and isolate, we can manage how the disease progresses through society, with a (very very slight, now) chance of stopping it. While not scientific, this is illustrative and well done that shows many aspects of disease spread: [VIDEO]

Part of 3 Blue 1 Brown's excellent latest video simulates travel restrictions and suggests they're not very effective: [VIDEO] . However, that was only for cutting travel between all communities by a factor of 4, rather than completely banning travel between two particular communities / countries.

Homemade masks made of a double layer of tea towel cloth are not only protecting others but also protect yourself nearly as effectively as N95 masks. The original research was made by PHE - Public Health England and the results are: B atrophaeus - ~1um : Surgical mask 96.35% filtering Double layer tea towel 96.71% filtering Bacteriophage MS2 - 23nm : Surgical mask 89.52% filtering Tea towel one layer 72.46% filtering The full paper is on researchgate, title: "Testing the Efficacy of Homemade Masks: Would They Protect in an Influenza Pandemic?" Also every super market should have the use of hand sanitizer obligatory at the entrance. There is a pretty good video for simulations of the epidemic spreading [1] - it is not going to stop until we either have a vaccine or ~60% of the population have antibodies. So the choice is to sit the next 12-18months at home or we fight the transmission of the virus. Masks cost nothing compared to the economic disaster that we are running into. The virus spreads via mouth and nose and secondary via the hands. So just stop it at the source. [1] [VIDEO]

This seems like such a no-brainer. 3Blue 1Brown did a video on epidemic simulation, and central markets, places where people in the region congregate, are major factors in the the rate of disease spread. Limiting the transfer coefficient in these places has the key to reducing this impact, assuming you can't prevent people from going there. [VIDEO]

Basically, some simulations show [1] if lots of people from an otherwise-socially-distancing population all go to one central location, that increases the R0 (i.e. spread) of the virus. This is a problem for anything that draws together lots of people from a large area into the same place - trade shows, sports events, music festivals, cinemas, conventions, busses and trains, food stores, beaches, universities and schools. We almost certainly haven't hit on the most optimal solution - in my country we closed schools, yet keeping crowded metro trains running? - but it wouldn't have been wise to wait for more data to come in before taking any action. [1] [VIDEO]

> ...noticeable reduction in the spread? I believe the SIR model [0] answers those questions. 3blue1brown's video on the topic is super digestible and informative [1]. In short, the discovery of a prophylactic if not a therapeutic treatment are our best immediate bets, absent which, quarantines and lockdowns will have to persist to curb the spread of a disease as infectious as covid-19 whilst we patiently wait for an effective vaccine against it [2]. [0] [LINK] [1] [VIDEO] [2] [LINK]

Garrett, thank you. * * * My inspiration for writing this came from watching 3Blue1Brown's YouTube video, "Simulating an epidemic": [VIDEO] His simulations cover several scenarios: - Simulating social distancing. - Having people travel to a central location (e.g. a grocery store). - What if the government is able to identify and isolate cases? - Simulating multiple communities with travel, that put quarantine measures in place. - What happens when people get tired of social distancing. * * * Thanks for sharing your code. I see that you are using gnuplot to visualize the output of the script. Have you considered using the csv module ( [LINK] ) to output the data, instead of using the print() function? That way, you wouldn't have to manually add in spaces around the data and stringify the numeric data. >>> import csv, sys >>> writer = csv.writer(sys.stdout, delimiter=" ") >>> writer.writerow(["#", "T", "S", "I", "R"]) # T S I R In the script, usage would look something like this: import csv import sys ... writer = csv.writer(sys.stdout, delimiter=" ") writer.writerow(["#", "T", "S", "I", "R"]) writer.writerow([T, S, I, R]) while I > 1000: ... writer.writerow([T, S, I, R])

There is a lengthier interview by UnHeard with Professor Johan Giesecke (another experienced epidemiologist) of Sweden on YouTube [0]. It’s worthwhile listening to their perspective. One of the comments he made was that no lockdown would work in democratic countries and the real test will be fatality rates between Sweden and others in 1 year. He believed it’ll be very little difference between those countries that have implement the lockdown and those that haven’t I had watched the simulation of covid19 infection by 1blue3brown[1] and some of his modeling appeared to me (maybe I misunderstood them) to confirm Swede’s point of view. Ultimately the same number of people got infected albeit in a “flatter” rate. However one counter point by Johan Giesecke was that these flattening the curve is motivated by the assumption of a static capacity of ER rooms to deal with this pandemic and he pointed out they have increased capacity. Overall I found listening to his side very informative, at least it gave me a different view of looking at from their view. [0] [LINK] [1] [VIDEO]

> screening everyone every day (which is very different from simply doing "more testing"). That seems like the logical conclusion of "more testing" to me. If we could, why wouldn't we? 3blue1brown on YouTube did an analysis which similarly shows that fast quarantining is the best way to mitigate the virus: [VIDEO]

It's pretty apparent the writers don't understand epidemiology themselves. What is this field, really? It's not medicine. Its most famous practitioners often don't have any medical training or background, e.g. Prof Ferguson who did his PhD in theoretical physics. The assumptions it uses are of the level of medical complexity that a small child can understand - essentially that people who spend time close to each other infect each other. The models don't take as input the molecular biology or DNA of a virus that's being simulated. Arguably it's not science. A basic characteristic of science is that it tells us something new about the world, something that can be proven true. Experiments are fundamental to science for that reason. Epidemiology is about the development of models. Models aren't science because they cannot ever tell us something truly new or unexpected: a model is ultimately just a rendering of its author's assumptions. When someone puts some formulas into R and copies the results into a paper, the model is acting as a kind of fancy PowerPoint. It's useful to illustrate the conclusions that follow from the assumptions, but it no more lets us discover new things than PowerPoint does. (n.b. Under this definition, a lot of what gets published as computer science also isn't science but rather engineering. "Data science" on the other hand is a tautology. That's a fine conclusion and one I'll happily defend.) It often gets described as mathematics. It's not mathematics

the whole point of the app is to isolate people that have been in contact with positive cases before they become contagious. if everyone has the app, you won't have an exponential growth. [VIDEO]

Not exactly a citation[1], but this agent-based simulation suggest that mitigations like quarantine are highly sensitive to how thoroughly they are applied: [VIDEO] [1] I'm putting this firmly in the "all models are wrong, but some models are useful" camp

BTW, he's currently co-teaching a course on Julia with some really great lectures on computational thinking: [LINK] He also had a nice video about epidemics/pandemics: [VIDEO] All using manim for visualizations.

> Technocrats used complicated mathematical models to determine that mass vaccination would create a "herd immunity" to disease. The SIR model is very simple [LINK] At the beginning of the covid19 epidemic everyone and their dog was writing one and publishing a demo version. It was interesting/scary to modify the parameters and see how thing changed. Also, try to fix misinformation in the models, like a 50% death rate. Anyway, a SIR model can be programed in 20 lines of codes, and is pretty straightforward. Adding a vaccine is like 5 additional lines at most. Also, the main idea is very simple to understand. (Someone should write a tutorial of SIR in Excel/GoogleSheets, it's interesting to see the graphics.) The most simple models just assume that everyone meets everyone randomly. It's more complicated if you want to consider a more realistic model of how people move. For a decision about lockdown you need the complex model, but to understand herd immunity the simple model is good enough. For a nice explanation of more complicated models, there is a video of 3Blue1Brown: "Simulating an epidemic" [VIDEO] Note that he explains that the models in this video are not 100% accurate, some number are chosen to get nice graphics, they are only useful to get some insight but not get quantitative predictions.

Dynamics of the virus spread is quite interesting. 3Blue1Brown made a fascinating video about how an epidemic spreads: [VIDEO]

I don’t get it. How can social distancing NOT help during a pandemic? Quote from the book 1491: “One reason is that Indians were fresh territory for many plagues, not just one. Smallpox, typhoid, bubonic plague, influenza, mumps, measles, whooping cough—all rained down on the Americas in the century after Columbus. (Cholera, malaria, and scarlet fever came later.) Having little experience with epidemic diseases, Indians had no knowledge of how to combat them. In contrast, Europeans were well versed in the brutal logic of quarantine. They boarded up houses in which plague appeared and fled to the countryside.” And here’s 3Blue1Brown running a simulation: [VIDEO]

OT - 3b1b did an interesting video on the epidemic modeling [0] that shaped my thinking a bit on the lockdown and caused me to lose a few friends to argue against the prolonged lockdowns. I believe he used the same technique deployed by major media institutions and health organizations. Basically the interesting conclusion was unless you apply a very rigorous lockdown (CCP style) the virus ultimately spreads. All you’re doing is spreading the time. As in “flattening the curve”. As some might recall the initial lockdown was proposed as a way to flatten the curve and allow the medical system to cope better by having a more gradual influx of patients. Which makes me think everyone promoting the initial lockdown understood its limitations and its objectives. However, for what ever reason “control demolition” of the economy to make Trump lose the election or who knows why it turned into a long term lockdown. I know this is a controversial topic, but it shouldn’t be. This is a clear case that the science of pandemic modeling suggested one path and for political reasons we were pushed to follow a course that caused a lot of damage, some irreversible, to the society. [0] [VIDEO]