sdaasdas
This commit is contained in:
parent
b4cc21a6b3
commit
dc6b1abcea
6 changed files with 776 additions and 605 deletions
151
index.html
151
index.html
|
@ -22,25 +22,21 @@
|
|||
</div>
|
||||
</div>
|
||||
|
||||
<p>"The only thing to fear is fear itself" is stupid.<sup id="fnref1"><a href="#fn1" rel="footnote">1</a></sup></p>
|
||||
<p>"The only thing to fear is fear itself" was stupid advice.</p>
|
||||
|
||||
<p>Sure, don't hoard toilet paper. But if someone's so scared to think about scary things, that they deny danger <em>when it's already here</em>, then they've got more problems<sup id="fnref2"><a href="#fn2" rel="footnote">2</a></sup> than toilet paper.</p>
|
||||
<p>If people fear fear itself, they'll deny danger because they don't want to create "mass panic". The problem's not fear, but how we <em>use</em> our fear. Fear, used well, gives you energy to deal with current dangers, and prepare for future dangers.</p>
|
||||
|
||||
<p>The problem's not fear, but how we <em>use</em> our fear. Taiwan and South Korea <em>bravely used their fear</em> (from SARS) to invest in "pandemic insurance", and it paid off in controlling COVID-19! Fear gives you energy to deal with present dangers & plan for future dangers – <em>if</em> you know how to channel your fear.</p>
|
||||
|
||||
<p>So, we (Marcel & Nicky) have channeled our COVID-19 fears into making these playable simulations – so that <em>you</em> can channel <em>your</em> fear into gaining a deep, intuitive understanding of:</p>
|
||||
<p>Honestly, the two of us (Marcel, epidemiologist + Nicky, artist/coder) are worried about the future. We bet you are, too. That's why we want to channel <em>our</em> worries into making these <strong>playable simulations</strong>, so that you can channel <em>your</em> worries into understanding:</p>
|
||||
|
||||
<ul>
|
||||
<li><strong>The Last Few Months</strong> (epidemiology 101, SIR model, R0 & Rt)</li>
|
||||
<li><strong>The Last Few Months</strong> (epidemiology 101, SEIR model, R & R<sub>0</sub>)</li>
|
||||
<li><strong>The Next Few Months</strong> (lockdowns, contact tracing, masks)</li>
|
||||
<li><strong>The Next Few Years</strong> (vaccines, summers, loss of immunity)</li>
|
||||
<li><strong>The Next Few Years</strong> (vaccines, loss of immunity?)</li>
|
||||
</ul>
|
||||
|
||||
<p>Note: We're publishing this on April 30th, 2020. Still the early days. As humanity learns more about COVID-19, our plans will and <em>should</em> change – but we hope this post will address 90%+ of all future possibilities!</p>
|
||||
<p>This guide is meant to give you hope <em>and</em> fear. To beat this virus <strong>in a way that also protects our mental & financial health</strong>, we need optimism to create plans, and pessimism to create backup plans. As Gladys Bronwyn Stern once said, <em>“The optimist invents the airplane and the pessimist the parachute.”</em></p>
|
||||
|
||||
<p>Honestly, some of the possibilities are scary. And some are hopeful! But preparing for the scary possibilites is what <em>creates</em> the hopeful possibilites. You don't get to save the prince/ss without facing the dragon.</p>
|
||||
|
||||
<p>Let's bravely use our fear, and face this dragon.</p>
|
||||
<p>So, buckle in: we're about to experience some turbulence.</p>
|
||||
|
||||
<div class="section">
|
||||
<div>
|
||||
|
@ -48,71 +44,57 @@
|
|||
</div>
|
||||
</div>
|
||||
|
||||
<p>...has been a real worldwide cram-school in Epidemiology 101.</p>
|
||||
<p>Pilots use flight simulators to learn how not to crash planes.</p>
|
||||
|
||||
<p>Pilots use flight simulators to learn how not to crash planes. <strong>Epidemiologists use epidemic simulators to learn how not to crash humanity.</strong></p>
|
||||
<p><strong>Epidemiologists use epidemic simulators to learn how not to crash humanity.</strong></p>
|
||||
|
||||
<p>So, let's set up an epidemic "flight simulator"! First, we need some simulation rules.</p>
|
||||
|
||||
<p>Let's say you have some Infected (i) people and not-yet-infected Susceptible (s) people. One (i) infects a (s), those 2 (i) infect another 2 (s), those 4 (i) infect another 4 (s), and so on:</p>
|
||||
<p>So, let's create a very simple "epidemic flight simulator"! Here, we have some (i) Infectious people & some not-yet-infected (s) Susceptible people. (i)s turn (s)s into more (i)s:</p>
|
||||
|
||||
<p>// pic</p>
|
||||
|
||||
<p><em>On average</em>, COVID-19 jumps from an (i) to a (s) every 4 days.<a href="source">1</a> The average # of days it takes for an (i) to infect an (s) is called the <strong>"generation time"</strong><a href="serial%20interval">2-note</a>. (Click the gray circles for sources, and the blue squares for side-notes!)</p>
|
||||
<p>At the start of a COVID-19 outbreak, it's estimated that the virus jumps from an (i) to an (s) every 4 days.<sup id="fnref1"><a href="#fn1" rel="footnote">1</a></sup> (<em>On average.</em> Remember, there's lots of variation.)</p>
|
||||
|
||||
<p><em>Rule #1: The more (i)s there are, the faster (s)s become (i)s.</em></p>
|
||||
<p>Here's a simulation of a population with <em>just</em> 0.001% (i) and 99.999% (s), over 6 months. If we simulate "double every 4 days" <em>and nothing else</em>, what happens?</p>
|
||||
|
||||
<p>// pic - rule</p>
|
||||
|
||||
<p>If we simulate <em>just this rule and nothing else</em>, here's what it looks like over 3 months, starting with 99.9% (s) and just 0.1% (i):</p>
|
||||
|
||||
<p><strong>Click "Start" play the simulation! You can then change the "generation time", and see how that changes the simulation:</strong></p>
|
||||
<p><strong>Click "Start" to play the simulation! (Afterwards, you can re-play the simulation with different settings)</strong></p>
|
||||
|
||||
<div class="sim">
|
||||
<iframe src="sim?stage=01" width="800" height="540"></iframe>
|
||||
<iframe src="sim?stage=epi-1" width="800" height="540"></iframe>
|
||||
</div>
|
||||
|
||||
<p>Starts small ("it's just a flu"), then explodes ("oh right, flus don't break hospitals in rich countries"). This is the "J-shaped" <strong>exponential growth curve</strong>.</p>
|
||||
<p>This is the <strong>exponential growth curve.</strong> Starts small, then explodes. "Oh it's just a flu" to "Oh right, flus don't create <em>mass graves in rich cities</em>". </p>
|
||||
|
||||
<p>But this simulation is wrong. There are things that prevent an (i) from infecting someone else – like if that other person is <em>already</em> an (i):</p>
|
||||
<p>// pic - exponential double rice</p>
|
||||
|
||||
<p>But, this simulation is wrong. Exponential growth, thankfully, can't go on forever. One thing that stops a virus from spreading is if others <em>already</em> have the virus:</p>
|
||||
|
||||
<p>// pic - 100% spread, 50% spread, 0% spread</p>
|
||||
|
||||
<p><em>Rule #2: The fewer (s)s there are, the slower (s)s become (i)s.</em></p>
|
||||
<p><strong>The more (i)s there are, the faster (s)s become (i)s, but the fewer (s)s there are, the <em>slower</em> (s)s become (i)s.</strong></p>
|
||||
|
||||
<p>// pic - rule</p>
|
||||
|
||||
<p>Now, what happens if we simulate <em>both</em> these rules?</p>
|
||||
|
||||
<p><strong>Again, click Start to play the simulation!</strong></p>
|
||||
<p>Now, what happens if we simulate that?</p>
|
||||
|
||||
<p>// sim</p>
|
||||
|
||||
<p>Starts small, explodes, then slows down again. This is the "S-shaped" <strong>logistic growth curve.</strong></p>
|
||||
<p>This is the "S-shaped" <strong>logistic growth curve.</strong> Starts small, explodes, then slows down again.</p>
|
||||
|
||||
<p>Still, this simulation predicts 100% of people will get the virus, and even the most pessimistic COVID-19 simulations don't predict <em>that</em>. </p>
|
||||
<p>But, this simulation is <em>still</em> wrong. We're missing the fact that (i) Infectious people eventually stop being infectious, either by 1) recovering, 2) "recovering" with lung damage, or 3) dying.</p>
|
||||
|
||||
<p>What we're missing: You stop being infectious for COVID-19 when you recover... or die.</p>
|
||||
<p>For simplicity's sake, let's pretend that all (i) Infectious people become (r) Recovered. (r)s can't be infected again, and let's pretend – <em>for now!</em> – that they stay immune for life.</p>
|
||||
|
||||
<p>For the sake not making these simulations too depressing, let's only simulate Infected (i) becoming (r) Recovered. (The math works out the same.) And let's assume <em>(for now!!!)</em> that (r)s can't get infected again. So, new rule:</p>
|
||||
<p>When you're infected with COVID-19, it's estimated you stay (i) infectious for 12 days.<sup id="fnref2"><a href="#fn2" rel="footnote">2</a></sup> (Again, <em>on average.</em>)</p>
|
||||
|
||||
<p><em>Rule #3: (i)s eventually become (r)s.</em> </p>
|
||||
|
||||
<p>// pic - rule</p>
|
||||
|
||||
<p>Let's have (i)s become (r)s after 14 days, <em>on average</em>.<a href="technical%20notes">3-note</a> This means some (i)s will recover <em>before</em> 14 days, and some recover <em>after</em>! This is closer to real life.</p>
|
||||
|
||||
<p>To show <em>only</em> Rule #3, here's a simulation starting with 100% (i):</p>
|
||||
<p>Here's a simulation that starts with 100% (i). Most people recover after 12 days, then most of the remainder recover after another 12 days, then most of the remainder <em>of that remainder</em> recover after another 12 days, etc:</p>
|
||||
|
||||
<p>// sim</p>
|
||||
|
||||
<p>This is the "flipped-J-shaped" <strong>exponential decay curve.</strong></p>
|
||||
<p>This is the opposite of exponential growth, the <strong>exponential decay curve</strong>.</p>
|
||||
|
||||
<p>Now, what happens if you simulate all 3 rules at once? What happens when you combine an S-shaped logistic curve with a flipped-J exponential decay curve?</p>
|
||||
<p>Now, what happens if you combine this with the S-shaped logistic curve of infection?</p>
|
||||
|
||||
<p>// pic</p>
|
||||
|
||||
<p>Let's find out:</p>
|
||||
<p>Let's find out. Here's a simulation of an epidemic <em>with</em> recovery:</p>
|
||||
|
||||
<p>// sim</p>
|
||||
|
||||
|
@ -120,70 +102,84 @@
|
|||
|
||||
<p>// pic: 3 rules</p>
|
||||
|
||||
<p>This is the <strong>SIR Model</strong>, the <em>second</em>-most important idea in epidemiology.</p>
|
||||
<p>This is the the <strong>SIR Model</strong>, ((s) <strong>S</strong>usceptible → (i) <strong>I</strong>nfectious → (r) <strong>R</strong>ecovered) the second-most important idea in Epidemiology 101.</p>
|
||||
|
||||
<p><strong>NOTE:</strong> The simulations you've been hearing in the news are <em>far</em> more complex than the ones you're seeing here! But the sims you'll play with here reach the same general conclusions, even if missing the nuances.</p>
|
||||
<p>Note: The simulations that inform policy are <em>far</em> more sophisticated than this! But the SIR model can still help us understand a lot about COVID-19, even if missing the nuances.</p>
|
||||
|
||||
<p>One nuance you could add is the <strong>SIRS Model</strong>, where the final "S" also stands for (s) Susceptible – this is when people recover, are immune for a bit, <em>then lose that immunity and can be infected again.</em> (We'll consider this in the Next Few Years section)</p>
|
||||
<p>Actually, let's add one more nuance: before an (s) becomes an (i), they first become an (e) Exposed person, when they're infect<em>ed</em> but not yet infect<em>ious</em> – they have the virus but can't pass it on (yet).</p>
|
||||
|
||||
<p>Another nuanced version is the <strong>SEIR Model</strong>, where the "E" stands for (e) Exposed, a brief period of time <em>after</em> you've been infected, but <em>before</em> you can infect others. This is called the <strong>"latent period"</strong>, and for COVID-19 it's around 3 days.<a href="">4</a></p>
|
||||
<p>(This variant is called the <strong>SEIR Model</strong>, where "E" stands for (e) Exposed. Note this <em>isn't</em> the everyday meaning of "exposed", where you might or might not have the virus. In this technical definition, "Exposed" means you definitely have it. Yeah, science terminology is bad.)</p>
|
||||
|
||||
<p>Here's what happens if you simulate that:</p>
|
||||
<p>For COVID-19, it's estimated that you're in this "latent period" for around 3 days.<sup id="fnref3"><a href="#fn3" rel="footnote">3</a></sup> What happens if we add that to the simulation?</p>
|
||||
|
||||
<p>// sim</p>
|
||||
|
||||
<p>Doesn't change much, so let's stick to the vanilla SIR model. We brought (e)s up because the exact timing of contagiousness is important in "contact tracing", which we'll explain in the Next Few Months section.</p>
|
||||
<p>Not much, actually! The "latent period" only changes <em>when</em> the peak happens, but the <em>height</em> of the peak – and total people infected – remain the same:</p>
|
||||
|
||||
<p>Oh! But almost forgot, the <em>first</em>-most important idea in epidemiology:</p>
|
||||
<p>// pics</p>
|
||||
|
||||
<p><strong>"R"</strong></p>
|
||||
<p>Why's that? Because of the <em>first</em>-most important idea in Epidemiology 101:</p>
|
||||
|
||||
<p>Which is short for "Reproduction Number". It's the <em>average</em> number of people an (i) will infect <em>before</em> they recover:</p>
|
||||
<p>// pic - <strong>"R"</strong></p>
|
||||
|
||||
<p>// pic - R>1 R=1 R<1</p>
|
||||
<p>Which is short for "Reproduction Number". It's the <em>average</em> number of people an (i) infects <em>before</em> they recover (or die).</p>
|
||||
|
||||
<p><strong>R0</strong> (pronounced R-nought) is the Reproduction Number for a virus <em>at the very beginning of an outbreak, before we have immunity or interventions</em>. (Also called "Basic Reproduction Number")</p>
|
||||
<p>// R > 1, R = 1, R < 1 pic</p>
|
||||
|
||||
<p><strong>Rt</strong> (the 't' stands for time) is the Reproduction Number <em>right now</em>, after we have some immunity or interventions. (Also called "Re", e standing for "Effective Reproduction Number". Also called just "R", to... confuse people)</p>
|
||||
<p><strong>R</strong> changes over the course of an outbreak, as we get more immunity & interventions.</p>
|
||||
|
||||
<p>// pic of R0 and Rt over time for the Famous Curve – with peak for inflection!</p>
|
||||
<p><strong>R<sub>0</sub></strong> (pronounced R-nought) is what R is <em>at the start of an outbreak, before immunity or interventions</em>. R<sub>0</sub> is also called the "basic reproduction number". R<sub>0</sub> more closely reflects the power of the virus itself, but it still changes from place to place. For example, because heat 'kills' coronaviruses, R<sub>0</sub> for COVID-19 is lower in hot places than cold ones. Not low enough to contain it, though.</p>
|
||||
|
||||
<p>(A lot of news outlets confuse these two Rs! They're different!)</p>
|
||||
<p>(A lot of news outlets – and even academic papers! – confuse R and R<sub>0</sub>. Again, science terminology is bad.)</p>
|
||||
|
||||
<p>The R0 for the flu<a href="more">6</a> is around 1.3. The R0 for COVID-19 is somewhere between 2 and 5.<a href="source">7</a> The huge uncertainty is because R0 depends on exactly how quickly new people are infected ("generation time") vs how quickly people recover<a href="technical%20note">8</a>:</p>
|
||||
<p>The R<sub>0</sub> for the flu<sup id="fnref4"><a href="#fn4" rel="footnote">4</a></sup> is around 1.3. The R<sub>0</sub> estimates for COVID-19 are usually between 2 and 3, maybe as high as 6.<sup id="fnref5"><a href="#fn5" rel="footnote">5</a></sup></p>
|
||||
|
||||
<p>In our simulations, an (i) recovers in 12 days, but infects one new (s) every 4 days. That means, <em>on average</em>, an (i) infects 3 (s)s before they recover. So for our simulations, R<sub>0</sub> is 3.</p>
|
||||
|
||||
<p><strong>Play around with this R<sub>0</sub> calculator, to see how R<sub>0</sub> depends on recovery time & new-infection time:</strong></p>
|
||||
|
||||
<p>// calc</p>
|
||||
|
||||
<p>But remember, the fewer (s)s there are, the <em>slower</em> (s)s become (i)s.
|
||||
R depends not just on R<sub>0</sub>, but also how many people are no longer Susceptible – due to, say, having recovered & gotten natural immunity.</p>
|
||||
|
||||
<p>// calc 2</p>
|
||||
|
||||
<p>When enough people have natural immunity, R < 1, and the virus is contained! This is called <strong>herd immunity</strong>, and while it's <em>terrible</em> policy, (we'll explain why later – it's not for the reason you may think!) it's essential to understanding Epidemiology 101.</p>
|
||||
|
||||
<p>Now, let's play the last simulation again, but showing R<sub>0</sub>, R over time, and the herd immunity threshold:</p>
|
||||
|
||||
<p>// sim</p>
|
||||
|
||||
<p>Rt for COVID-19 depends on the interventions we do (or don't) have, as well as how many people <em>aren't</em> (s) Susceptible. (because they're (r) Recovered, currently (i) Infected, or... dead.)</p>
|
||||
<p>Note: Total cases (the gray curve) does not stop at herd immunity, but <em>overshoots</em> it! And it does this <em>exactly when</em> current cases (the pink curve) peaks. This happens no matter how you change the settings:</p>
|
||||
|
||||
<p>// sim</p>
|
||||
<p>// pic</p>
|
||||
|
||||
<p>Note that when (s)% is low enough, you can get Rt<1 – <em>containing the virus!</em> This is called <strong>the "herd immunity" threshold</strong>. "Herd immunity" is a terrible <em>policy</em> (TODO: explain why), but it's important for understanding epidemiology.</p>
|
||||
<p>This is because, by definition, when there are more non-(s)s than the herd immunity threshold, you get R < 1. And, by definition, R < 1 means new cases stop growing.</p>
|
||||
|
||||
<p>Now, let's run the same SIR model simulation again, but this time showing 1) Rt changing over time, and 2) the herd immunity threshold:</p>
|
||||
<p>If there's only one lesson you take away from this whole guide, here it is, in big shiny letters:</p>
|
||||
|
||||
<p>// sim</p>
|
||||
<h1 id="toc_3">R > 1 = bad</h1>
|
||||
|
||||
<p>Note how total cases ((i)+(r)) <em>overshoots</em> the herd immunity threshold! And the <em>exact</em> moment it does this is when infections peak <em>and</em> when Rt drops below 1!</p>
|
||||
<h1 id="toc_4">R < 1 = good (R=1, meh)</h1>
|
||||
|
||||
<p>If there's only one lesson you take away today, here it is, in big shiny letters:</p>
|
||||
<p><strong>This means: we do NOT need to catch all transmissions, or even nearly all transmissions, to stop COVID-19!</strong></p>
|
||||
|
||||
<h1 id="toc_3">Rt>1 = bad</h1>
|
||||
<p>It's a paradox. COVID-19 is incredibly contagious, yet to contain it, we "only" need to stop 67% of infections. 67%?! If that was a school grade, that's a D+. But if R<sub>0</sub> = 3, cutting that by 67% gives us R = 0.99, which is R < 1, which means the virus is contained!</p>
|
||||
|
||||
<h1 id="toc_4">Rt<1 = good</h1>
|
||||
<p>(And even if, extreme-worst-case, R<sub>0</sub> = <em>6</em>, you still "only" need to stop 84% of transmissions. That's a B grade.)</p>
|
||||
|
||||
<p><strong>NOTE: We do not need to catch all transmissions, or even nearly all transmissions, to stop COVID-19.</strong></p>
|
||||
|
||||
<p>It's a paradox – COVID-19 is incredibly contagious, yet to contain it, we "only" need to stop 72% of infections. 72%?! That's, like, a C– grade. But if R0 = 3.5, then reducing that by 72% will make Rt < 1 = good.</p>
|
||||
|
||||
<p>(And even if worst-case, R0=5, you "only" need to stop 80%. That's a B–.)</p>
|
||||
<p>// calculator - custom</p>
|
||||
|
||||
<p><em>Every</em> COVID-19 intervention you've heard of – handwashing, social distancing, lockdowns, self-isolation, contact tracing & quarantining, face masks, even "herd immunity" – they're <em>all</em> doing the same thing:</p>
|
||||
|
||||
<p>Reducing Rt.</p>
|
||||
<p>Getting R < 1.</p>
|
||||
|
||||
<p>Let's see how we can get Rt<1 – in a way that protects not just our physical health, but also our mental health, social health, <em>and</em> financial health!</p>
|
||||
<p>So now, let's use our "epidemic flight simulator" to figure out the next few months! How will we get R < 1 in a way that protects not just our physical health, <strong>but also our mental health, social health, <em>and</em> financial health?</strong></p>
|
||||
|
||||
<p>Brace yourselves for an emergency landing...</p>
|
||||
|
||||
<!--
|
||||
<div class="section">
|
||||
<div>
|
||||
<h1 id="toc_5">The Next Few Months</h1>
|
||||
|
@ -559,6 +555,7 @@
|
|||
<hr>
|
||||
|
||||
<p>(TODO: US vs Korea/Taiwain resources)</p>
|
||||
-->
|
||||
|
||||
<div class="footnotes">
|
||||
<hr>
|
||||
|
|
528
words.md
528
words.md
|
@ -1,528 +0,0 @@
|
|||
# What Happens Next?
|
||||
|
||||
## COVID-19 Possibilities, Explained With Playable Simulations
|
||||
|
||||
"The only thing to fear is fear itself" is stupid.[^1]
|
||||
|
||||
[^1]: hello!
|
||||
|
||||
Sure, don't hoard toilet paper. But if someone's so scared to think about scary things, that they deny danger *when it's already here*, then they've got more problems[^2] than toilet paper.
|
||||
|
||||
[^2]: but a snitch ain't one
|
||||
|
||||
The problem's not fear, but how we *use* our fear. Taiwan and South Korea *bravely used their fear* (from SARS) to invest in "pandemic insurance", and it paid off in controlling COVID-19! Fear gives you energy to deal with present dangers & plan for future dangers – *if* you know how to channel your fear.
|
||||
|
||||
So, we (Marcel & Nicky) have channeled our COVID-19 fears into making these playable simulations – so that *you* can channel *your* fear into gaining a deep, intuitive understanding of:
|
||||
|
||||
* **The Last Few Months** (epidemiology 101, SIR model, R0 & Rt)
|
||||
* **The Next Few Months** (lockdowns, contact tracing, masks)
|
||||
* **The Next Few Years** (vaccines, summers, loss of immunity)
|
||||
|
||||
Note: We're publishing this on April 30th, 2020. Still the early days. As humanity learns more about COVID-19, our plans will and *should* change – but we hope this post will address 90%+ of all future possibilities!
|
||||
|
||||
Honestly, some of the possibilities are scary. And some are hopeful! But preparing for the scary possibilites is what *creates* the hopeful possibilites. You don't get to save the prince/ss without facing the dragon.
|
||||
|
||||
Let's bravely use our fear, and face this dragon.
|
||||
|
||||
---
|
||||
|
||||
# The Last Few Months
|
||||
|
||||
...has been a real worldwide cram-school in Epidemiology 101.
|
||||
|
||||
Pilots use flight simulators to learn how not to crash planes. **Epidemiologists use epidemic simulators to learn how not to crash humanity.**
|
||||
|
||||
So, let's set up an epidemic "flight simulator"! First, we need some simulation rules.
|
||||
|
||||
Let's say you have some Infected (i) people and not-yet-infected Susceptible (s) people. One (i) infects a (s), those 2 (i) infect another 2 (s), those 4 (i) infect another 4 (s), and so on:
|
||||
|
||||
// pic
|
||||
|
||||
*On average*, COVID-19 jumps from an (i) to a (s) every 4 days.[1](source) The average # of days it takes for an (i) to infect an (s) is called the **"generation time"**[2-note](serial interval). (Click the gray circles for sources, and the blue squares for side-notes!)
|
||||
|
||||
*Rule #1: The more (i)s there are, the faster (s)s become (i)s.*
|
||||
|
||||
// pic - rule
|
||||
|
||||
If we simulate *just this rule and nothing else*, here's what it looks like over 3 months, starting with 99.9% (s) and just 0.1% (i):
|
||||
|
||||
**Click "Start" play the simulation! You can then change the "generation time", and see how that changes the simulation:**
|
||||
|
||||
// sim
|
||||
|
||||
Starts small ("it's just a flu"), then explodes ("oh right, flus don't break hospitals in rich countries"). This is the "J-shaped" **exponential growth curve**.
|
||||
|
||||
But this simulation is wrong. There are things that prevent an (i) from infecting someone else – like if that other person is *already* an (i):
|
||||
|
||||
// pic - 100% spread, 50% spread, 0% spread
|
||||
|
||||
*Rule #2: The fewer (s)s there are, the slower (s)s become (i)s.*
|
||||
|
||||
// pic - rule
|
||||
|
||||
Now, what happens if we simulate *both* these rules?
|
||||
|
||||
**Again, click Start to play the simulation!**
|
||||
|
||||
// sim
|
||||
|
||||
Starts small, explodes, then slows down again. This is the "S-shaped" **logistic growth curve.**
|
||||
|
||||
Still, this simulation predicts 100% of people will get the virus, and even the most pessimistic COVID-19 simulations don't predict *that*.
|
||||
|
||||
What we're missing: You stop being infectious for COVID-19 when you recover... or die.
|
||||
|
||||
For the sake not making these simulations too depressing, let's only simulate Infected (i) becoming (r) Recovered. (The math works out the same.) And let's assume *(for now!!!)* that (r)s can't get infected again. So, new rule:
|
||||
|
||||
*Rule #3: (i)s eventually become (r)s.*
|
||||
|
||||
// pic - rule
|
||||
|
||||
Let's have (i)s become (r)s after 14 days, *on average*.[3-note](technical notes) This means some (i)s will recover *before* 14 days, and some recover *after*! This is closer to real life.
|
||||
|
||||
To show *only* Rule #3, here's a simulation starting with 100% (i):
|
||||
|
||||
// sim
|
||||
|
||||
This is the "flipped-J-shaped" **exponential decay curve.**
|
||||
|
||||
Now, what happens if you simulate all 3 rules at once? What happens when you combine an S-shaped logistic curve with a flipped-J exponential decay curve?
|
||||
|
||||
// pic
|
||||
|
||||
Let's find out:
|
||||
|
||||
// sim
|
||||
|
||||
And *that's* where that famous curve comes from! It's not a bell curve, it's not even a "log-normal" curve. It has no name. But you've seen it a zillion times, and beseeched to flatten.
|
||||
|
||||
// pic: 3 rules
|
||||
|
||||
This is the **SIR Model**, the *second*-most important idea in epidemiology.
|
||||
|
||||
**NOTE:** The simulations you've been hearing in the news are *far* more complex than the ones you're seeing here! But the sims you'll play with here reach the same general conclusions, even if missing the nuances.
|
||||
|
||||
One nuance you could add is the **SIRS Model**, where the final "S" also stands for (s) Susceptible – this is when people recover, are immune for a bit, *then lose that immunity and can be infected again.* (We'll consider this in the Next Few Years section)
|
||||
|
||||
Another nuanced version is the **SEIR Model**, where the "E" stands for (e) Exposed, a brief period of time *after* you've been infected, but *before* you can infect others. This is called the **"latent period"**, and for COVID-19 it's around 3 days.[4]()
|
||||
|
||||
Here's what happens if you simulate that:
|
||||
|
||||
// sim
|
||||
|
||||
Doesn't change much, so let's stick to the vanilla SIR model. We brought (e)s up because the exact timing of contagiousness is important in "contact tracing", which we'll explain in the Next Few Months section.
|
||||
|
||||
Oh! But almost forgot, the *first*-most important idea in epidemiology:
|
||||
|
||||
**"R"**
|
||||
|
||||
Which is short for "Reproduction Number". It's the *average* number of people an (i) will infect *before* they recover:
|
||||
|
||||
// pic - R>1 R=1 R<1
|
||||
|
||||
**R0** (pronounced R-nought) is the Reproduction Number for a virus *at the very beginning of an outbreak, before we have immunity or interventions*. (Also called "Basic Reproduction Number")
|
||||
|
||||
**Rt** (the 't' stands for time) is the Reproduction Number *right now*, after we have some immunity or interventions. (Also called "Re", e standing for "Effective Reproduction Number". Also called just "R", to... confuse people)
|
||||
|
||||
// pic of R0 and Rt over time for the Famous Curve – with peak for inflection!
|
||||
|
||||
(A lot of news outlets confuse these two Rs! They're different!)
|
||||
|
||||
The R0 for the flu[6](more) is around 1.3. The R0 for COVID-19 is somewhere between 2 and 5.[7](source) The huge uncertainty is because R0 depends on exactly how quickly new people are infected ("generation time") vs how quickly people recover[8](technical note):
|
||||
|
||||
// sim
|
||||
|
||||
Rt for COVID-19 depends on the interventions we do (or don't) have, as well as how many people *aren't* (s) Susceptible. (because they're (r) Recovered, currently (i) Infected, or... dead.)
|
||||
|
||||
// sim
|
||||
|
||||
Note that when (s)% is low enough, you can get Rt<1 – *containing the virus!* This is called **the "herd immunity" threshold**. "Herd immunity" is a terrible *policy* (TODO: explain why), but it's important for understanding epidemiology.
|
||||
|
||||
Now, let's run the same SIR model simulation again, but this time showing 1) Rt changing over time, and 2) the herd immunity threshold:
|
||||
|
||||
// sim
|
||||
|
||||
Note how total cases ((i)+(r)) *overshoots* the herd immunity threshold! And the *exact* moment it does this is when infections peak *and* when Rt drops below 1!
|
||||
|
||||
If there's only one lesson you take away today, here it is, in big shiny letters:
|
||||
|
||||
# Rt>1 = bad
|
||||
# Rt<1 = good
|
||||
|
||||
**NOTE: We do not need to catch all transmissions, or even nearly all transmissions, to stop COVID-19.**
|
||||
|
||||
It's a paradox – COVID-19 is incredibly contagious, yet to contain it, we "only" need to stop 72% of infections. 72%?! That's, like, a C– grade. But if R0 = 3.5, then reducing that by 72% will make Rt < 1 = good.
|
||||
|
||||
(And even if worst-case, R0=5, you "only" need to stop 80%. That's a B–.)
|
||||
|
||||
*Every* COVID-19 intervention you've heard of – handwashing, social distancing, lockdowns, self-isolation, contact tracing & quarantining, face masks, even "herd immunity" – they're *all* doing the same thing:
|
||||
|
||||
Reducing Rt.
|
||||
|
||||
Let's see how we can get Rt<1 – in a way that protects not just our physical health, but also our mental health, social health, *and* financial health!
|
||||
|
||||
---
|
||||
|
||||
# The Next Few Months
|
||||
|
||||
...could have been worse.
|
||||
|
||||
###Scenario 0: Do Absolutely Nothing
|
||||
|
||||
For COVID-19, 1 in 20 (i)s need to be hospitalized. In rich countries like the US and UK, there's 1 hospital bed for every 1000 people. Therefore: a rich country can handle a maximum of 20 (i)s per 1000 people – or, a maximum of 2% of the population being simultaneously sick.
|
||||
|
||||
Here's the same simulation from before, but with the "2%" threshold drawn:
|
||||
|
||||
// sim
|
||||
|
||||
It's not good.
|
||||
|
||||
That's the same thing the March 16th Imperial College report found: if we do nothing, hospitals break. Almost everyone gets infected. Even with a low 0.5% infection fatality ratio, 80% of people infected in a large country like the US still means over a million dead... *IF* we did nothing.
|
||||
|
||||
(A lot of news & social media chose to report the scary bit, *without* "IF WE DO NOTHING". Fear was channeled into clicks, not understanding. *Sigh.*)
|
||||
|
||||
###Scenario 1: Flatten The Curve
|
||||
|
||||
Handwashing was discovered in ____ by the doctor _______, when he realized that by getting his staff to wash their hands, child deaths in his hospital were cut by *90%!*
|
||||
|
||||
Doctors around the world immediately hailed his life-saving discovery, and ha ha just kidding they committed him to an asylum where he was beat to death by guards.
|
||||
|
||||
In any case, frequent handwashing reduces your chances of catching influenza by 50%![9]() And if we combine this with other hygiene tips – cough into your elbow, don't touch your face – let's guess-timate that 100% compliance (which we will *NOT* get) will result in a 60% reduction in new infections, in Rt:
|
||||
|
||||
// controls
|
||||
|
||||
It can't get Rt<1, but it *does* reduce it! How does that affect the epidemic?
|
||||
|
||||
// sim
|
||||
|
||||
That's a... *better* catastrophe.
|
||||
|
||||
Contrary to many news & social media posts, "flattening the curve" *does also reduce total cases*. But as long as Rt is still above 1, our hospitals will still most likely shatter.
|
||||
|
||||
That's what the Imperial College report also found: any attempt at mere **"mitigation"** (Reduce Rt, but still Rt>1 = bad) will fail, and the only way out is **"suppression"**. (Reduce Rt, so that Rt<1 = good!)
|
||||
|
||||
*Crush* the curve, not just flatten it. For example, by doing a...
|
||||
|
||||
###Scenario 2: Months-Long Lockdown (we are here)
|
||||
|
||||
There's different degrees of "physical distancing". (previously called "social distancing") At the mildest, avoiding crowds. At the strongest, a full city-wide lockdown.
|
||||
|
||||
London's full lockdown reduced Rt by 70%.[11]() So, let's guess-timate that as the maximum effect for distancing.
|
||||
|
||||
Here's how hygiene & distancing together change Rt:
|
||||
|
||||
// calc
|
||||
|
||||
That's Rt<1 = good!
|
||||
|
||||
Let's see what happens if we *crush* the curve with a lockdown for 3 months, then finally, *finally* return to normal life:
|
||||
|
||||
**Remember, you can re-play the simulation, and change the sliders *WHILE* it's running, to simulate your own COVID-19 strategy! You can also pause & slow down the simulation:**
|
||||
|
||||
// sim
|
||||
|
||||
Oh.
|
||||
|
||||
Right, as soon as you remove the lockdown, Rt>1 again, and so you get a spike in cases that's almost as bad as if you'd done *nothing at all.*
|
||||
|
||||
**A lockdown isn't a cure, it's just a restart.**
|
||||
|
||||
So, what, do we just lockdown again & again?
|
||||
|
||||
###Scenario 3: Intermittent Lockdown
|
||||
|
||||
// sim
|
||||
|
||||
This was one solution suggested by the March 16 Imperial College report, and analyzed again by Marc Lipsitch ______ etc. [https://science.sciencemag.org/content/early/2020/04/14/science.abb5793?]
|
||||
|
||||
This *would* keep hospitals below capacity! You just have to... shut everything down for 2 months, every 3 months, until a vaccine is available in 18 months. That's... one year total out of 18 months.
|
||||
|
||||
Look, it's all well & good to draw a line on a graph saying "healthcare capacity", but there's lots of important things we *can't* simulate here. Like:
|
||||
|
||||
Mental Health) Loneliness is one of the biggest risk factors for depression, anxiety, and suicide. And it's as negatively associated with an early death as smoking 15 cigarettes a day.
|
||||
|
||||
Financial Health) "What about the economy" sounds like you care more about dollars than lives, but "the economy" isn't just stocks: it's people's ability to provide food & shelter for their loved ones, to invest in their kids' futures, and enjoy arts, foods, videogames – the stuff makes life worth living. And besides, poverty *itself* has horrible impacts on mental and physical health.
|
||||
|
||||
Not saying we should rule out intermittent lockdowns! But it's not ideal.
|
||||
|
||||
Wait, didn't we say Taiwan & South Korea "bravely used their fear" to control COVID-19? For 4 whole months? How?
|
||||
|
||||
###Scenario 4: Test, Trace, Isolate
|
||||
|
||||
You may be thinking:
|
||||
|
||||
*Sure, we \*could\* have done what Taiwan + South Korea did at the start, but it's too late now. We missed the start.*
|
||||
|
||||
But that's exactly it! A lockdown isn't a cure, it's just a restart – **and a fresh start is what we need.** (TODO: Actually, South Korea started late!)
|
||||
|
||||
The lockdown will let us reduce (i) cases, and buy time to copy what Taiwan & South Korea are already successfully doing: isolating COVID-19 cases, *and finding out who've they been in extended close contact with* (**"contact tracing"**) *and quarantining them too*.
|
||||
|
||||
(Pedantic note: **"isolate"** is for infected cases, **"quarantine"** is for contacts)
|
||||
|
||||
Why do we need to quarantine the contacts? Because they could have been (e) Exposed & caught the virus, but not know it yet:
|
||||
|
||||
// timeline
|
||||
|
||||
If you *only* isolate the cases, the virus can still spread:
|
||||
|
||||
// timeline
|
||||
|
||||
But if you *also* quarantine the contacts, you stop the spread, by staying one step ahead!
|
||||
|
||||
// timeline
|
||||
|
||||
(TODO: 30 min+ exposure)
|
||||
|
||||
Contact tracing was how they contained Ebola in (where?) Africa! And that was just good ol' fashioned "ask people who they met" contact tracing.
|
||||
|
||||
...which, unfortunately, will not work for COVID-19. Interviews are too slow and human memory is too unreliable. [MARCEL'S SOURCE] There's only *3 days* between being exposed to the virus (e) and being able to infect others (i).
|
||||
|
||||
So, regrettably, some countries have resorted to privacy-invasive techniques, like grabbing loads of citizens' phone location data. But does protecting human lives mean surrendering to Big Brother?
|
||||
|
||||
HECK NO
|
||||
|
||||
**Here's a short comic we made, explaining how you can do digital contact tracing in a privacy-protecting way.** And when we say "privacy-protecting", we mean that *even if the central server was hacked and all its data stolen*, the hacker would learn *nothing* about people's identities, locations, or who met who.
|
||||
|
||||
(And here's a 3Blue1Brown video adaptation of our comic! Thanks Grant!)
|
||||
|
||||
And this isn't just "in theory". Several apps are *already* being developed for this. The European council vote (FILL IN). And Google/Apple's new announcement specifically supports the privacy-protecting protocol as described in our comic above. (Don't trust Google/Apple? Neither do we! The beauty of the protocol is that *it doesn't rely on trust.*)
|
||||
|
||||
Okay okay, enough tooting our own horn. How does isolating cases & quarantining contacts reduce Rt?
|
||||
|
||||
University of Oxford study estimates that:
|
||||
|
||||
* (i) Symptomatic **cases** account for 40% of new infections. So by isolating cases, **you reduce Rt by up to 40%.**
|
||||
* (e) Pre-symptomatic & A-symptomatic **contacts** account for 50%. So by quarantining contacts, **you can reduce Rt by up to 50%.**
|
||||
* Stuff like doorknobs accounts for the rest, 10%.
|
||||
|
||||
[Loong note about "pre" vs "a" symptomatic & how the media screwed it up AGAIN]()
|
||||
|
||||
So, combined, isolating cases & quarantining contacts can get Rt comfortably below 1, *even with NO physical distancing!*
|
||||
|
||||
// calc
|
||||
|
||||
Remember: **we do not need to catch all transmissions, or even nearly all transmissions, to stop COVID-19.** So the fact that not everybody is able (or willing) to download a privacy-protecting contact tracing app isn't a dealbreaker.
|
||||
|
||||
We don't need to catch *all* contacts, isolate *all* cases, or even wash *all* the hands. Just enough to get that C– grade of 72%, to get Rt<1 = good.
|
||||
|
||||
(do wash your hands, though)
|
||||
|
||||
Alright, enough chat. Here's a simulation of using a lockdown as reset, then switching to "Test, Trace, Isolate":
|
||||
|
||||
// sim
|
||||
|
||||
And here it is again, with a vaccine at 18 months, which converts (s) into an immune (r), without having to become a (i). This gives us "herd immunity" the *right* way, and we can *finally* stop all other interventions.
|
||||
|
||||
(actually, keep washing your hands. come on, a doctor was beaten to death in an asylum.)
|
||||
|
||||
// sim
|
||||
|
||||
So that's it!
|
||||
|
||||
That's currently the best working plan, recommended by several independent teams of epidemiologists & policymakers from across the political spectrum. (LINKS) Lockdown to get a fresh start, switch to Taiwan & South Korea's strategy later.
|
||||
|
||||
But...
|
||||
|
||||
...you may be feeling a knot in your stomach. Things have *already* gone horribly wrong, more stuff could *still* go horribly wrong with this plan, right?
|
||||
|
||||
You're dang right it could. Let's channel that fear... into making some *backup plans:*
|
||||
|
||||
###Scenario X: Other Interventions & Backup Plans
|
||||
|
||||
If handwashing + case isolation + contact quarantining *still* isn't enough to get Rt<1... we can supplement it with three things:
|
||||
|
||||
**Deep Cleaning:**
|
||||
|
||||
Remember we said "stuff like doorknobs" accounts for 10% of new infections? The technical jargon for things that can pass a virus from one human to another is a **"fomite".**
|
||||
|
||||
10% means frequent deep cleanings of public spaces – subways, libraries, and malls – can reduce Rt by up to 10%. Which sounds useless, but if it reduces Rt from 1.05 to 0.95... that's Rt<1 = lives saved.
|
||||
|
||||
// calc?
|
||||
|
||||
**Masks For All:**
|
||||
|
||||
[small brain] Correlation implies causation!
|
||||
|
||||
[normal brain] Correlation doesn't imply causation, you need Randomized Controlled Trials (RCTs) to prove things.
|
||||
|
||||
[large brain] Actually, under Bayes' Theorem, *all* correlations are evidence for causation, because the likelihood of {seeing a correlation, given causation} is greater than the likelihood of {seeing a correlation, given *no* causation}. It's just not 100% proof, because *nothing* in science is 100% proof, not even RCTs (hence the replication crisis). **Evidence isn't 0% or 100%, they have a full range of "weights".** And though correlational evidence has a lower "weight" than an RCT, it *is still evidence.* (See this 3Blue1Brown video for a visual explanation of Bayes' Theorem)
|
||||
|
||||
What we're trying to say is:
|
||||
|
||||
There aren't any RCTs (yet) testing "Cloth masks prevent COVID-19 spread" *specifically*. But there's lots of *suggestive* evidence, if of lower "weight":
|
||||
|
||||
* Staff in hospitals without masks are more likely to die of COVID-19
|
||||
* Taiwan & South Korea have widespread public wearing of masks
|
||||
* For colds & flus, masks reduce droplets/aerosols from the *wearer*.
|
||||
|
||||
Pandemics are like poker. Act only when you "have enough info", and you'll lose everything at stake. You'll never have enough info, just cost/benefit analyses under uncertainty. Like so:
|
||||
|
||||
Cost of cloth masks (certain): Small. Same as handwashing.
|
||||
|
||||
Benefit of cloth masks (uncertain): They probably don't stop *the wearer* from getting COVID-19, but they probably stop a pre-symptomatic wearer from *spreading* COVID-19. Let's guess masks reduce Rt by 0% to 20%. *Even though "0%" is still likely*, the average "expected value" is *halfway* between 0% and 20% – that is, 10%, same as deep cleaning, but at minuscule cost.
|
||||
|
||||
Analysis: If someone offered you a coin flip, where tails = nothing happens, and heads = 1000s of lives saved... and the price for playing this game is a rag and two rubber bands... even though "nothing" is as likely as "lives saved", you should do it.
|
||||
|
||||
Cloth masks for all: do it!
|
||||
|
||||
// calc?
|
||||
|
||||
**Summer:**
|
||||
|
||||
Okay, this is not an "intervention" we have control of, but it *does* help reduce Rt!
|
||||
|
||||
For every extra 1° Celsius (2.2° Fahrenheit), Rt drops by ___%. The average difference between winter & summer in New York is 15°C (60°F), so summer will make Rt drop by _%.
|
||||
|
||||
Many news sites (wrongly) report summer won't slow COVID-19. They're probably trying not to get your hopes up: with R0=3.5, a _% reduction is Rt=\_, still above 1.
|
||||
|
||||
But still, it's *something*. If we have limited resources, we can scale back some interventions in the summer – so we can scale them higher in the winter.
|
||||
|
||||
// calc? over time
|
||||
|
||||
**A "Circuit Breaker" Lockdown:**
|
||||
|
||||
And if all that *still* isn't enough to get Rt<1... we can do another lockdown.
|
||||
|
||||
But because Rt was reduced dramatically, we wouldn't have to do a 2-month-lockdown-every-3-months! Probably just *one* more 1-month lockdown, between now and when we have a vaccine.
|
||||
|
||||
Here's a simulation of that (with sliders for *ALL* the interventions):
|
||||
|
||||
// sim
|
||||
|
||||
. . .
|
||||
|
||||
We hope these plans give you hope.
|
||||
|
||||
It *is* possible to keep Rt<1, *without* locking down for most of 18 months. With plans like "Test, Trace, Isolate", supplemented with backup plans like "Masks For All", we can get back to a normal-ish life!
|
||||
|
||||
Sure, your hands may be dry. But you'll get to invite a date out to a comics bookstore! You'll get to watch the latest cash-grab Hollywood sequel with friends. You'll get to people-watch at a library, taking joy in people going about the simple business of *being alive.*
|
||||
|
||||
Life will go on, even under the worst-case scenario.
|
||||
|
||||
So now, let's use our fear's energy, and plan for some *even worse* worst-case scenarios:
|
||||
|
||||
---
|
||||
|
||||
# The Next Few Years
|
||||
|
||||
You get COVID-19, and recover. Or you get the COVID-19 vaccine! Either way, you're now immune...
|
||||
|
||||
...*for how long?*
|
||||
|
||||
SARS, which was closely related (TODO: is it?) to this new coronavirus, gave its survivors around 2 years of immunity.[12](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2851497/). Some coronaviruses, like the ones that cause "the" common cold[13], give you just [1 year of immunity](https://pubmed.ncbi.nlm.nih.gov/2170159/). (TODO: MERS' immunity)
|
||||
|
||||
Let's think about the scariest scenario: immunity doesn't last.
|
||||
|
||||
*Rule #4: (r)s eventually become (s)s*
|
||||
|
||||
// pic
|
||||
|
||||
The SIRS model: the (r) Recovered become (s) Susceptible again.
|
||||
|
||||
Let's simulate what that'll look like, with *no* interventions:
|
||||
|
||||
// sim
|
||||
|
||||
Previously, with no interventions, we only had *one* hospital-breaking spike. Now, we have several, *and* the simulation comes to a rest with % of (i) infected *permanently above* hospital capacity.
|
||||
|
||||
(If you replay the simulation above with immunity lasting 3 years, that wouldn't be so bad! The % of (i) would rest comfortably *below* capacity. There'd still be spikes, but you can deal with them using the same interventions listed in last section)
|
||||
|
||||
It's like a pendulum: total (i)+(r) cases swings around the "herd immunity" threshold, before settling exactly at "herd immunity", where Rt=1. The virus no longer grows or shrinks. It's just with us forever: it's **endemic.**
|
||||
|
||||
// pic?
|
||||
|
||||
Thankfully, summer will make it better by reducing Rt:
|
||||
|
||||
// sim
|
||||
|
||||
Oh wait no it doesn't. Summer *does* reduce new people becoming (i) infected, but that also reduces new people becoming (r) immune. Which means immunity in the population will drop *even further* with summer, allowing for *big regular spikes* in the winter.
|
||||
|
||||
It's like a pendulum where you're moving the top back and forth: that just makes the cycles *worse*.
|
||||
|
||||
// pic?
|
||||
|
||||
Finally, the *worst* worst-case:
|
||||
|
||||
What if, like HIV, there's just *never* a vaccine?
|
||||
|
||||
Our only option now is to increase our capacity for COVID-19 cases. You could do this directly, by creating more hospital beds and ventilators. Or you could do this indirectly, by creating treatments for COVID-19, so that if you *do* get it, you're less likely to need a hospital bed or ventilator.
|
||||
|
||||
Here's the same simulation, but 1) starting with herd immunity (which wanes quickly), and 2) with adjustable hospital capacity:
|
||||
|
||||
// sim
|
||||
|
||||
HIV/AIDS killed millions, mostly in marginalized communities. And yet, despite it being the worst-case pandemic scenario, and despite all the stigma against people who have it... HIV isn't a death sentence anymore.
|
||||
|
||||
HIV has no vaccine. There's *definitely* no herd immunity. And yet, with treatments like antiretroviral therapy, people can and *are* living full lives with the virus. COVID-19 is devastating, but nowhere as much as HIV.
|
||||
|
||||
Life will go on, even under the *worst* worst-case scenario.
|
||||
|
||||
...
|
||||
|
||||
That said, the virus behind COVID-19 is way simpler than HIV, so there'll almost definitely be a vaccine, even if it only grants immunity for a year. If so, we'll just have to do a vaccination campaign each autumn – and we can just do this alongside our regular flu shots:
|
||||
|
||||
// sim
|
||||
|
||||
**Finally, here's a Simulation Sandbox, with *every* option available. You can now also share your *own* simulations!**
|
||||
|
||||
// sim
|
||||
|
||||
Play around to intuitively understand the core rules of epidemiology.
|
||||
|
||||
Try simulating different COVID-19 scenarios, plans, and backup plans.
|
||||
|
||||
Ask questions, try to find an answer with the sim, and share your sim with others.
|
||||
|
||||
This (again, very *VERY* basic!) simulation has let us answer so many questions about the past few months, next few months, and next few years.
|
||||
|
||||
So now, let's return to...
|
||||
|
||||
---
|
||||
|
||||
# The Now
|
||||
|
||||
In summary, here's how we bravely use our fear, slay the dragon, and save the lives of millions of princes(ses):
|
||||
|
||||
**PHASE 1) Lockdown to get a fresh start.**
|
||||
|
||||
Get current (i)s low, while building capability to do...
|
||||
|
||||
**PHASE 2) "Test, Trace, Isolate"**
|
||||
|
||||
We replace lockdown with other ways to get Rt<1. Life gets back to normal-ish! 🎉
|
||||
|
||||
More testing so we can *actually* tell what Rt currently is.
|
||||
|
||||
Create policies to get cases to isolate/quarantine. Paid leave & bonus financial incentives if they do, *maybe* fines if they don't.
|
||||
|
||||
Use *privacy-protecting* contact tracing apps to find contacts. Remember, not everybody has to have the app to get Rt<1.
|
||||
|
||||
If Rt still not below 1: "Masks For All". Get most people to wear at *least* cloth face masks.
|
||||
|
||||
If Rt *still* not below 1: Deep clean public spaces often. Mild social distancing. Maybe one or two more "circuit breaker" lockdowns. (but still avoiding "lockdown for most of 18 months"!)
|
||||
|
||||
This will buy us time to finally do...
|
||||
|
||||
**PHASE 3) Vaccinate!**
|
||||
|
||||
If immunity doesn't last long: Vaccination campaign every autumn, like we already do for flu shots.
|
||||
|
||||
If vaccine is *never* available: Raise our capacity for COVID-19 cases by creating more hospital beds & ventilators, and developing antivirals & treatments. (which we should be doing *anyway!*)
|
||||
|
||||
**What's this mean for YOU, _right now?_**
|
||||
|
||||
**For everyone:** Respect the lockdown so we can get out of Phase I asap. Keep washing those hands. Make your own masks. Download a *privacy-protecting* contact tracing app when those are available next month. Stay healthy, physically & mentally! And write your local policymaker to get off their butt and...
|
||||
|
||||
**For policymakers:** Create policies that compensate (or reward!) folks who have to self-isolate/quarantine. Direct funds into all the stuff we should be building, like...
|
||||
|
||||
**For builders:** Build tests. Build ventilators. Build masks – cloth, surgical and N95. Build apps. Build antivirals and other treatments. Build vaccines. Build science.
|
||||
|
||||
Will we *need* all that? "Probably" not, the same way you "probably" won't need safety belts, fire insurance, or parachutes on planes. It's like doing a cost/benefit analysis of Russian Roulette: the chance of disaster is small, but the *cost* of disaster is far, far bigger.
|
||||
|
||||
In situations like this, it pays to listen *honestly* to your fears. Don't deny or downplay them, just face them, and prepare for them.
|
||||
|
||||
The only thing to fear is people who think the only thing to fear is fear itself.
|
||||
|
||||
---
|
||||
|
||||
(TODO: US vs Korea/Taiwain resources)
|
498
words/words.md
Normal file
498
words/words.md
Normal file
|
@ -0,0 +1,498 @@
|
|||
# What Happens Next?
|
||||
|
||||
## COVID-19 Futures, Explained With Playable Simulations
|
||||
|
||||
"The only thing to fear is fear itself" was stupid advice.
|
||||
|
||||
If people fear fear itself, they'll deny danger because they don't want to create "mass panic". The problem's not fear, but how we *use* our fear. Fear, used well, gives you energy to deal with current dangers, and prepare for future dangers.
|
||||
|
||||
Honestly, the two of us (Marcel, epidemiologist + Nicky, artist/coder) are worried about the future. We bet you are, too. That's why we've channeled *our* worries into making these **playable simulations**, so that you can channel *your* worries into understanding:
|
||||
|
||||
* **The Last Few Months** (epidemiology 101, SEIR model, R & R<sub>0</sub>)
|
||||
* **The Next Few Months** (lockdowns, contact tracing, masks)
|
||||
* **The Next Few Years** (loss of immunity? no safe vaccine?)
|
||||
|
||||
This guide is meant to give you hope *and* fear. To beat this virus **in a way that also protects our mental & financial health**, we need optimism to create plans, and pessimism to create backup plans. As Gladys Bronwyn Stern once said, *“The optimist invents the airplane and the pessimist the parachute.”*
|
||||
|
||||
So, buckle in: we're about to experience some turbulence.
|
||||
|
||||
---
|
||||
|
||||
# The Last Few Months
|
||||
|
||||
Pilots use flight simulators to learn how not to crash planes.
|
||||
|
||||
**Epidemiologists use epidemic simulators to learn how not to crash humanity.**
|
||||
|
||||
So, let's create a very simple "epidemic flight simulator"! Here, we have some (i) Infectious people & some not-yet-infected (s) Susceptible people. (i)s turn (s)s into more (i)s:
|
||||
|
||||
// pic
|
||||
|
||||
At the start of a COVID-19 outbreak, it's estimated that the virus jumps from an (i) to an (s) every 4 days.[^1] (*On average.* Remember, there's lots of variation.)
|
||||
|
||||
[^1]: source
|
||||
|
||||
Here's a simulation of a population with *just* 0.001% (i) and 99.999% (s), over 6 months. If we simulate "double every 4 days" *and nothing else*, what happens?
|
||||
|
||||
**Click "Start" to play the simulation! (Afterwards, you can re-play the simulation with different settings)**
|
||||
|
||||
// sim
|
||||
|
||||
This is the **exponential growth curve.** Starts small, then explodes. "Oh it's just a flu" to "Oh right, flus don't create *mass graves in rich cities*".
|
||||
|
||||
// pic - exponential double rice
|
||||
|
||||
But, this simulation is wrong. Exponential growth, thankfully, can't go on forever. One thing that stops a virus from spreading is if others *already* have the virus:
|
||||
|
||||
// pic - 100% spread, 50% spread, 0% spread
|
||||
|
||||
**The more (i)s there are, the faster (s)s become (i)s, but the fewer (s)s there are, the *slower* (s)s become (i)s.**
|
||||
|
||||
Now, what happens if we simulate that?
|
||||
|
||||
// sim
|
||||
|
||||
This is the "S-shaped" **logistic growth curve.** Starts small, explodes, then slows down again.
|
||||
|
||||
But, this simulation is *still* wrong. We're missing the fact that (i) Infectious people eventually stop being infectious, either by 1) recovering, 2) "recovering" with lung damage, or 3) dying.
|
||||
|
||||
For simplicity's sake, let's pretend that all (i) Infectious people become (r) Recovered. (r)s can't be infected again, and let's pretend – *for now!* – that they stay immune for life.
|
||||
|
||||
When you're infected with COVID-19, it's estimated you stay (i) infectious for 12 days.[^2] (Again, *on average.*)
|
||||
|
||||
[^2]: source
|
||||
|
||||
Here's a simulation that starts with 100% (i). Most people recover after 12 days, then most of the remainder recover after another 12 days, then most of the remainder *of that remainder* recover after another 12 days, etc:
|
||||
|
||||
// sim
|
||||
|
||||
This is the opposite of exponential growth, the **exponential decay curve**.
|
||||
|
||||
Now, what happens if you combine this with the S-shaped logistic curve of infection?
|
||||
|
||||
// pic
|
||||
|
||||
Let's find out. Here's a simulation of an epidemic *with* recovery:
|
||||
|
||||
// sim
|
||||
|
||||
And *that's* where that famous curve comes from! It's not a bell curve, it's not even a "log-normal" curve. It has no name. But you've seen it a zillion times, and beseeched to flatten.
|
||||
|
||||
// pic: 3 rules
|
||||
|
||||
This is the the **SIR Model**, ((s) **S**usceptible → (i) **I**nfectious → (r) **R**ecovered) the second-most important idea in Epidemiology 101.
|
||||
|
||||
Note: The simulations that inform policy are *far* more sophisticated than this! But the SIR model can still help us understand a lot about COVID-19, even if missing the nuances.
|
||||
|
||||
Actually, let's add one more nuance: before an (s) becomes an (i), they first become an (e) Exposed person, when they're infect*ed* but not yet infect*ious* – they have the virus but can't pass it on (yet).
|
||||
|
||||
(This variant is called the **SEIR Model**, where "E" stands for (e) Exposed. Note this *isn't* the everyday meaning of "exposed", where you might or might not have the virus. In this technical definition, "Exposed" means you definitely have it. Yeah, science terminology is bad.)
|
||||
|
||||
For COVID-19, it's estimated that you're in this "latent period" for around 3 days.[^3] What happens if we add that to the simulation?
|
||||
|
||||
[^3]: source
|
||||
|
||||
// sim
|
||||
|
||||
Not much, actually! The "latent period" only changes *when* the peak happens, but the *height* of the peak – and total people infected – remain the same:
|
||||
|
||||
// pics
|
||||
|
||||
Why's that? Because of the *first*-most important idea in Epidemiology 101:
|
||||
|
||||
// pic - **"R"**
|
||||
|
||||
Which is short for "Reproduction Number". It's the *average* number of people an (i) infects *before* they recover (or die).
|
||||
|
||||
// R > 1, R = 1, R < 1 pic
|
||||
|
||||
**R** changes over the course of an outbreak, as we get more immunity & interventions.
|
||||
|
||||
**R<sub>0</sub>** (pronounced R-nought) is what R is *at the start of an outbreak, before immunity or interventions*. R<sub>0</sub> is also called the "basic reproduction number". <!--R<sub>0</sub> more closely reflects the power of the virus itself, but it still changes from place to place. For example, because heat 'kills' coronaviruses, R<sub>0</sub> for COVID-19 is lower in hot places than cold ones. Not low enough to contain it, though. [source!]-->
|
||||
|
||||
(A lot of news outlets – and even academic papers! – confuse R and R<sub>0</sub>. Again, science terminology is bad.)
|
||||
|
||||
The R<sub>0</sub> for the flu[^r0_flu] is around 1.3. The R<sub>0</sub> estimates for COVID-19 are usually between 2 and 3, maybe as high as 6.[^r0_covid]
|
||||
|
||||
[^r0_flu]: source
|
||||
|
||||
[^r0_covid]: source
|
||||
|
||||
In our simulations, an (i) recovers in 12 days, but infects one new (s) every 4 days. That means, *on average*, an (i) infects 3 (s)s before they recover. So for our simulations, R<sub>0</sub> is 3.
|
||||
|
||||
**Play around with this R<sub>0</sub> calculator, to see how R<sub>0</sub> depends on recovery time & new-infection time:**
|
||||
|
||||
// calc
|
||||
|
||||
But remember, the fewer (s)s there are, the *slower* (s)s become (i)s.
|
||||
R depends not just on R<sub>0</sub>, but also how many people are no longer Susceptible – due to, say, having recovered & gotten natural immunity.
|
||||
|
||||
// calc 2
|
||||
|
||||
When enough people have natural immunity, R < 1, and the virus is contained! This is called **herd immunity**, and while it's *terrible* policy (we'll explain why later – it's not for the reason you may think!), it's essential to understanding Epidemiology 101.
|
||||
|
||||
Now, let's play the last simulation again, but showing R<sub>0</sub>, R over time, and the herd immunity threshold:
|
||||
|
||||
// sim
|
||||
|
||||
Note: Total cases (the gray curve) does not stop at herd immunity, but *overshoots* it! And it does this *exactly when* current cases (the pink curve) peaks. This happens no matter how you change the settings:
|
||||
|
||||
// pic
|
||||
|
||||
This is because, by definition, when there are more non-(s)s than the herd immunity threshold, you get R < 1. And, by definition, R < 1 means new cases stop growing.
|
||||
|
||||
If there's only one lesson you take away from this whole guide, here it is, in big shiny letters:
|
||||
|
||||
# R > 1 = bad
|
||||
# R < 1 = good (R=1, meh)
|
||||
|
||||
**This means: we do NOT need to catch all transmissions, or even nearly all transmissions, to stop COVID-19!**
|
||||
|
||||
It's a paradox. COVID-19 is incredibly contagious, yet to contain it, we "only" need to stop 67% of infections. 67%?! If that was a school grade, that's a D+. But if R<sub>0</sub> = 3, cutting that by 67% gives us R = 0.99, which is R < 1, which means the virus is contained!
|
||||
|
||||
// pic calc
|
||||
|
||||
*Every* COVID-19 intervention you've heard of – handwashing, social distancing, lockdowns, self-isolation, contact tracing & quarantining, face masks, even "herd immunity" – they're *all* doing the same thing:
|
||||
|
||||
Getting R < 1.
|
||||
|
||||
So now, let's use our "epidemic flight simulator" to figure out the next few months! How will we get R < 1 in a way that protects not just our physical health, **but also our mental health, social health, *and* financial health?**
|
||||
|
||||
Brace yourselves for an emergency landing...
|
||||
|
||||
---
|
||||
|
||||
# The Next Few Months
|
||||
|
||||
...could have been worse. Here's a parallel universe we avoided:
|
||||
|
||||
###Scenario 0: Do Absolutely Nothing
|
||||
|
||||
Around 1 in 20 people (i) infected with COVID-19 need an ICU (Intensive Care Unit).[^ic] In a rich country like the US, there's around 1 ICU per 3000 people.[^icus] Therefore, the US can handle 20 out of 3000 people being simultaneously (i) infected with COVID-19, or, 0.67% of the population.
|
||||
|
||||
Even if we *tripled* that capacity, to handle 2% of the population simultaneously (i) infected, here's what would happen *if we had done absolutely nothing:*
|
||||
|
||||
// sim
|
||||
|
||||
It's not good.
|
||||
|
||||
That's what the March 16 Imperial College report found: if we did absolutely nothing, we run out of ICUs & 80%+ of the population gets infected.
|
||||
|
||||
Even if only 0.5% of (i)s die – a generous assumption when there's no more ICUs – in a large country like the US, with 300 million people, 0.5% of 80% of 300 million = still 1.2 million dead, *IF WE DID NOTHING.*
|
||||
|
||||
(A lot of news outlets reported the scary bit, *without* "IF WE DO NOTHING". One can channel fear into good use, but here, fear was channeled into clicks. *Sigh.*)
|
||||
|
||||
###Scenario 1: Flatten The Curve / Herd Immunity
|
||||
|
||||
The "Flatten The Curve" plan was touted by every public health organization, while the United Kingdom's original "herd immunity" plan was universally booed. They were *the same plan.*[^yong] The UK just communicated their plan terribly.
|
||||
|
||||
[^yong]: s
|
||||
|
||||
Both plans, though, are horribly flawed.
|
||||
|
||||
First, the plans' specific interventions: mainly, handwashing & social distancing.
|
||||
|
||||
Increased handwashing cuts flus & colds in "developed nations" by 25%[^handwashing], while a city-wide lockdown cuts close person-to-person proximity by 70%[^london]. So, let's assume handwashing can reduce R by *up to* 25%, and distancing can reduce R by *up to* 70%:
|
||||
|
||||
[^handwashing]: s
|
||||
|
||||
[^london]: s
|
||||
|
||||
// calc
|
||||
|
||||
Now, let's simulate what happens if we apply fervent handwashing and *mild* social distancing – so that R is lower, but still above 1:
|
||||
|
||||
// sim
|
||||
|
||||
Three notes:
|
||||
|
||||
1. This *reduces* total cases! Lots of folks think "Flattening The Curve" *spread outs* cases without reducing the total. This is impossible in *any* Epidemiology 101 model. But because the news reported "80%+ of world will be infected" as inevitable – not *IF WE DO NOTHING* – folks thought total cases will be the same no matter what. *Sigh.*
|
||||
|
||||
2. Due to the extra interventions, current cases ((e)+(i)) peaks *before* herd immunity is reached. And in fact, total cases doesn't overshoot, but *goes to* herd immunity – the UK's plan! At that point, R < 1, you can let go of all other interventions, and COVID-19 stays contained! Well, except for one problem...
|
||||
|
||||
3. You still run out of ICUs. Several times over. For months.
|
||||
|
||||
That was the other finding of the March 16 Imperial College report, which convinced the UK to abandon its original plan. Any attempt at **mitigation** (reduce R, but R > 1) will fail. The only way out is **suppression** (reduce R so that R < 1).
|
||||
|
||||
// pic: difference
|
||||
|
||||
That is, don't merely "flatten" the curve, *crush* the curve. For example, with a...
|
||||
|
||||
###Scenario 2: Months-Long Lockdown (we are here)
|
||||
|
||||
Let's see what happens if we *crush* the curve with a lockdown for 5 months. Get R < 1, and smack (i)s to the bottom.
|
||||
|
||||
Then finally, *finally*, we can return to normal life:
|
||||
|
||||
// sim
|
||||
|
||||
Oh.
|
||||
|
||||
Right, this is the "second wave" everyone's talking about. As soon as we remove the lockdown, we get R > 1 again. So, a single leftover (i) (or a single imported (i)) can cause a spike in cases that's almost as bad as if we'd done Scenario 0: Absolutely Nothing.
|
||||
|
||||
**A lockdown isn't a cure, it's just a restart.**
|
||||
|
||||
So, what, do we just lockdown again & again?
|
||||
|
||||
###Scenario 3: Intermittent Lockdown
|
||||
|
||||
This solution was first suggested by the Imperial College, and later again by Harvard[^lockdown_harvard]:
|
||||
|
||||
[^lockdown_harvard]: https://science.sciencemag.org/content/early/2020/04/14/science.abb5793?
|
||||
|
||||
// sim
|
||||
|
||||
This *would* in fact keep cases below ICU capacity! We'd just need to... shut everything down for 2 months, open up for 1 month, then repeat until a vaccine is available in 18 months. That's a year in total.
|
||||
|
||||
(And if there's no vaccine, repeat until herd immunity is reached... in 2022.)
|
||||
|
||||
Look, it's nice to draw a line saying "healthcare capacity", but there's lots of important things we *can't* simulate here. Like:
|
||||
|
||||
**Mental Health:** Loneliness is one of the biggest risk factors for depression, anxiety, and suicide. And it's as negatively associated with an early death as smoking 15 cigarettes a day.
|
||||
|
||||
**Financial Health:** "What about the economy" sounds like you care more about dollars than lives, but "the economy" isn't just stocks: it's people's ability to provide food & shelter for their loved ones, to invest in their kids' futures, and enjoy arts, foods, videogames – the stuff makes life worth living. And besides, poverty *itself* has horrible impacts on mental and physical health.
|
||||
|
||||
But wait... haven't Taiwan, South Korea, and other countries in East Asia *already* contained COVID-19? For 4 whole months? Without a single country-wide lockdown?
|
||||
|
||||
How?
|
||||
|
||||
###Scenario 4: Test, Trace, Isolate
|
||||
|
||||
You may be thinking:
|
||||
|
||||
*Sure, we \*could've\* done what Taiwan + South Korea did at the start, but it's too late now. We missed the start.*
|
||||
|
||||
But that's exactly it! “A lockdown isn't a cure, it's just a restart”... **and a fresh start is what we need.**
|
||||
|
||||
To understand how Taiwan & South Korea have contained COVID-19, we need to understand the exact timeline of the virus:
|
||||
|
||||
// timeline
|
||||
|
||||
The problem is, if cases self-isolate *only* when they know they're infected (showing symptoms), the virus can still spread:
|
||||
|
||||
// timeline
|
||||
|
||||
But if you can find who the case had recent close contact to... (e.g. was within 6 feet of someone for 30+ minutes in the last 14 days) ...and quarantine them as well, you stop the spread, by staying one step ahead!
|
||||
|
||||
// timeline
|
||||
|
||||
This is called **contact tracing**. It's a core part of Taiwan & South Korea's strategies, and several teams of Western epidemiologists + economists + policymakers, from across the political spectrum, have converged on the same answer: *this is what we need to control COVID-19, while protecting our mental & financial health.*
|
||||
|
||||
Which is why, in the coming months, you may be asked to install a "contact tracing app" on your phone. Which sounds like a privacy nightmare, but teams of epidemiologists and cryptographers **have already created *anonymous, decentralized* contact tracing apps.**
|
||||
|
||||
(Here's a comic we made about how!)
|
||||
|
||||
But, critics say, contact tracing apps won't catch *all* transmissions. Some people don't have smartphones. Some transmissions are through surfaces like doorknobs.
|
||||
|
||||
True, *and it doesn't matter*. We don't *need* to catch all or even nearly all transmissions! Just enough to get R < 1.
|
||||
|
||||
Specifically, a University of Oxford study estimates that isolating symptomatic cases can reduce R by up to 40%, and that quarantining their pre/a-symptomatic contacts can reduce R by up to 50%:
|
||||
|
||||
(p.s: a rant about "pre" vs "a"-symptomatic)
|
||||
|
||||
// calc
|
||||
|
||||
So even if you don't isolate *all* cases or quarantine *all* contacts, you can still get R below 1, *with no social distancing!*
|
||||
|
||||
Okay, enough talk. Here's a simulation of:
|
||||
|
||||
1. A few-month lockdown, followed by...
|
||||
2. A switch to "Test, Trace, Isolate", for 18 months until...
|
||||
3. We can vaccinate folks, turning (s)s to (r)s directly, which gets us "herd immunity" the *right* way, which means R < 1, which means...
|
||||
4. We win.
|
||||
|
||||
// sim
|
||||
|
||||
Note: "Test, Trace, Isolate" is only possible when we have low enough *current* cases (which a lockdown will do) and high enough test kits (which a lockdown will buy us time to do). We also recommend that policymakers create policies giving people paid [see Vi's thing]
|
||||
|
||||
So that's it! That's how we make an emergency landing on this plane.
|
||||
|
||||
That's how we win.
|
||||
|
||||
...
|
||||
|
||||
But... you may be feeling a knot in your stomach. Things have *already* gone wrong, what if this plan goes wrong too? That's good – that's fear, which gives us energy to come up with *backup plans*.
|
||||
|
||||
The pessimist invents the parachute.
|
||||
|
||||
###Scenario 4+: Masks For All, Summer, Circuit Breakers
|
||||
|
||||
What if R<sub>0</sub> is way higher than we thought, and the above interventions *still* aren't enough to get R < 1?
|
||||
|
||||
If so, here's a few supplements:
|
||||
|
||||
**Masks For All:**
|
||||
|
||||
If there's a surgical mask shortage in your country, obviously don't hoard surgical masks. But should we all wear *homemade* cloth masks? Or make enough surgical masks so *everyone* can wear one?
|
||||
|
||||
"Wait", you might ask, "I thought scientists showed masks don't stop you from getting sick?"
|
||||
|
||||
You're right. Masks don't stop you from getting sick – they stop you from getting *others* sick. For colds/flus, surgical masks don't block incoming aerosols[^incoming]... but they block of 70% of *outgoing* aerosols[^outgoing aerosols], and almost 100% of *outgoing* droplets[^outgoing droplets]. (And home-made cloth masks are around half as efficient as surgical masks)
|
||||
|
||||
[^incoming]:
|
||||
|
||||
[^outgoing aerosols]:
|
||||
|
||||
[^outgoing droplets]:
|
||||
|
||||
Since pre-symptomatic folks account for almost *half* of transmissions, that's a big deal!
|
||||
|
||||
// pic: why masks don't protect you, but protect others
|
||||
|
||||
Still, there aren't (yet) any experiments of masks for the public on COVID-19 *specifically*. However:
|
||||
|
||||
1) There also aren't any experiments where we push people out of planes, and half of them get placebo parachutes. Sometimes, "solid" evidence isn't possible to get, so we have to settle for "circumstantial" evidence. (In this case: X, Y, Z) (sources)
|
||||
|
||||
2) Pandemics are like poker. Act only when you "have solid evidence", and you'll lose everything at stake. This isn't theory, it's practice – we *have* to make cost/benefit analyses under uncertainty. The cost of masks for the public (especially cloth masks) is low. The benefit is at worst nothing, at best a massive reduction in R.
|
||||
|
||||
How much, exactly? The "blocks 70% of areosols" study was for surgical masks, with proper fit. Assuming imperfect use by the public, let's guess surgical masks "only" reduce R by up to 50%. And since cloth masks are half as effective, let's guess cloth masks reduce R by up to 25%:
|
||||
|
||||
(If you think our guesses are too high – and that's good to challenge our assumptions! – just scale the sliders down)
|
||||
|
||||
// calc
|
||||
|
||||
Masks *alone* won't get R < 1. But if handwashing + "Test, Trace, Isolate" only gets us to R = 1.2, having just 2/3 of people wear *cloth* masks would tip that over to R < 1, virus contained!
|
||||
|
||||
**Summer:**
|
||||
|
||||
Okay, this isn't an "intervention" we have control of, but it will help! Some news outlets have reported that summer won't do anything to COVID-19. They're half right: summer won't get R < 1, but it *will* reduce R.
|
||||
|
||||
For every extra 1° Celsius (2.2° Fahrenheit), the R for COVID-19 drops by X%. The summer-winter difference in New York City is 15°C (60°F), so summer will make R drop by X%.
|
||||
|
||||
// calc, over time - in your country
|
||||
|
||||
It's not R < 1, but if we have limited resources, we can scale back some interventions in the summer – so we can scale them *higher* in the winter.
|
||||
|
||||
**A "Circuit Breaker" Lockdown:**
|
||||
|
||||
And if all that *still* isn't enough to get R < 1... we can do another lockdown.
|
||||
|
||||
But we wouldn't have to be 2-months-closed/1-month-open over & over! Because R is significantly reduced, we'd only need one or two more "circuit breaker" lockdowns before a vaccine is available. (Singapore had to do this recently, "despite" having controlled COVID-19 for 4 months. That's not failure: this *is* what success takes.)
|
||||
|
||||
Here's a simulation a "lazy case" scenario: we lockdown hard now, then do a moderate amount of "Test, Trace, Isolate" + a moderate amount of *cloth* "Masks For All", and do one circuit breaker lockdown before a vaccine is available:
|
||||
|
||||
// sim
|
||||
|
||||
. . .
|
||||
|
||||
We hope these plans give you hope.
|
||||
|
||||
It *is* possible to keep R < 1, *while protecting our mental, social, and financial health*. Use the lockdown as a restart, keep R < 1 with privacy-protecting contract tracing apps, supplemented with cloth masks for all... and life can get back to a normal-ish!
|
||||
|
||||
Sure, your hands may be dry. But you'll get to invite a date out to a comics bookstore! You'll get to go out with friends to watch the latest Hollywood cash-grab. You'll get to people-watch at a library, taking joy in people going about the simple business of *being alive.*
|
||||
|
||||
Even under the worst-case scenario... life perseveres.
|
||||
|
||||
So now, let's plan for some *even worse* worst-case scenarios.
|
||||
|
||||
Water landing, get your life jacket, and please follow the lights to the emergency exits:
|
||||
|
||||
---
|
||||
|
||||
# The Next Few Years
|
||||
|
||||
You get COVID-19, and recover. Or you get the COVID-19 vaccine. Either way, you're now immune...
|
||||
|
||||
...*for how long?*
|
||||
|
||||
The news reports folks who test positive again after testing negative, but those were most likely false positives/negatives. Still, the possibility of **waning immunity** is very real.
|
||||
|
||||
"Coronavirus" is a category of viruses.[^nitpick_1] The coronavirus responsible for COVID-19 is most closely related to the coronavirus responsible for SARS. SARS probably gave its survivors around 2 years of immunity.[^SARS immunity] The coronaviruses that cause "the"[^nitpick_2] common cold give you 1 year of immunity[^cold immunity]. So here's the scary question:
|
||||
|
||||
[^nitpick_1]: sas
|
||||
|
||||
[^nitpick_2]: adsa
|
||||
|
||||
[^SARS immunity]: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2851497/
|
||||
|
||||
[^cold immunity]: https://pubmed.ncbi.nlm.nih.gov/2170159/
|
||||
|
||||
What if COVID-19 immunity doesn't last?
|
||||
|
||||
Here's a simulation starting with 100% (i), exponentially decaying into (r)s after 12 days... but then back to susceptible, no-immunity (s)s after 1 year:
|
||||
|
||||
// sim
|
||||
|
||||
Now let's simulate a COVID-19 outbreak, over 10 years, with no interventions... *if immunity only lasts a year:*
|
||||
|
||||
// sim
|
||||
|
||||
This is the **SEIRS Model**. The final "S" stands for (s) Susceptible, again.
|
||||
|
||||
Previously, we only had *one* hospital-breaking spike. Now, we have several, *and* (i) cases come to a rest *permanently above* hospital capacity.
|
||||
|
||||
R = 1, it's **endemic.**
|
||||
|
||||
Thankfully, because summer reduces R, it'll make the situation better:
|
||||
|
||||
// sim
|
||||
|
||||
Counterintuitively, summer makes spikes *worse*, and repeated!
|
||||
|
||||
Summer reduces new (i)s, but that in turn reduces new (r)s. Which means immunity plummets in the summer, *creating* large spikes in the winter. And the spikes overwhelm ICUs over and over, *even if we double hospital capacity*.
|
||||
|
||||
Thankfully, the solution to this is pretty straightforward – just vaccinate people every fall/winter, like we do with flu shots:
|
||||
|
||||
// sim
|
||||
|
||||
But here's the scarier question:
|
||||
|
||||
What if there's just *never* a vaccine?
|
||||
|
||||
**To be clear: this is unlikely.** Sure, there's never been a vaccine for any of the other coronaviruses before, but that's because SARS was eradicated quickly, and "the" common cold wasn't worth the investment. Coronaviruses aren't any more complex than the viruses we already have vaccines for, so most public health scientists expect a vaccine *eventually*, even if years from now.
|
||||
|
||||
Still, there are valid worries: when we find a COVID-19 vaccine, what if we can't make enough? What if it's not *safe* enough?[^nature]
|
||||
|
||||
[^nature]: link
|
||||
|
||||
Even in this nightmare scenario, we still have 3 ways out. From most to least terrible:
|
||||
|
||||
1) Do the R < 1 interventions, but *looser*, so that we can get a steady supply of (i) to reach "natural herd immunity".
|
||||
|
||||
(Warning: this will result in many deaths & damaged lungs. *And* it won't work if immunity doesn't last.)
|
||||
|
||||
// sim
|
||||
|
||||
2) Do the R < 1 interventions forever. Contact tracing & wearing masks just becomes a new norm in the post-COVID-19 world, like how STI tests & wearing condoms became a new norm in the post-HIV world. (Nobody suggested "herd immunity" for HIV...)
|
||||
|
||||
3) Do the R < 1 interventions until we develop treatments that make COVID-19 way, way less likely to need hospitalization. (Which we should be doing *anyway!*) This is effectively the same as increasing our hospital capacity:
|
||||
|
||||
// sim
|
||||
|
||||
Even under the *worst* worst-case scenario... life perseveres.
|
||||
|
||||
But, you may have more questions, more intervention plans & scenarios you'd like to try. Great! **Here's a Simulation Sandbox Mode, with *every* option available! Simulate & play around to your heart's content:**
|
||||
|
||||
// sim
|
||||
|
||||
Even just Epidemiology 101 + a basic "epidemic flight simulator" can teach us so much. It's let us answer questions about the past few months, next few months, and next few years.
|
||||
|
||||
So now, let's return to...
|
||||
|
||||
---
|
||||
|
||||
# The Now
|
||||
|
||||
We've scrambled into the life rafts. It's time to find dry land.
|
||||
|
||||
Here's the growing consensus COVID-19 plan, with some less-consensus backup plans:
|
||||
|
||||
// pic
|
||||
|
||||
So what does this mean for YOU, right now?
|
||||
|
||||
**For everyone:** Respect the lockdown so we can get out of Phase I asap. Keep washing those hands. Make your own masks. Download a *privacy-protecting* contact tracing app when those are available next month. Stay healthy, physically & mentally! And write your local policymaker to get off their butt and...
|
||||
|
||||
**For policymakers:** Create policies that support folks who have to self-isolate/quarantine. Maybe make mask laws, if enough evidence accrues. Direct more funds into the stuff we should be building, like...
|
||||
|
||||
**For builders:** Build tests. Build ventilators. Build personal protective equipment for hospitals. Build masks. Build apps. Build antivirals, prophylactics, and other treatments that aren't vaccines. Build vaccines. Build better COVID-19 monitoring, so we can plan better. Build science. Build morale. Build hope.
|
||||
|
||||
Will we *need* all that? "Probably" not, the same way you "probably" won't need parachutes, life jackets, or inflatable rafts on an airplane. It's like a cost/benefit analysis of Russian Roulette: the chance of disaster may be small, but the *cost* of disaster is far, far worse.
|
||||
|
||||
So, don't deny or downplay fear to try to build hope. Our fear should *team up* with our hope, like the inventors of airplanes & parachutes. Preparing for horrible futures is how we *create* hopeful futures.
|
||||
|
||||
The only thing to fear is people who think the only thing to fear is fear itself.
|
||||
|
||||
// summary slides
|
||||
|
||||
// footnotes
|
204
words/words_epi_101.html
Normal file
204
words/words_epi_101.html
Normal file
|
@ -0,0 +1,204 @@
|
|||
<!DOCTYPE html>
|
||||
<html>
|
||||
|
||||
<head>
|
||||
|
||||
<meta charset="utf-8">
|
||||
<title>words_epi_101</title>
|
||||
|
||||
|
||||
|
||||
</head>
|
||||
|
||||
<body>
|
||||
|
||||
<h1 id="toc_0">What Happens Next?</h1>
|
||||
|
||||
<h2 id="toc_1">COVID-19 Futures, Explained With Playable Simulations</h2>
|
||||
|
||||
<p>"The only thing to fear is fear itself" was stupid advice.</p>
|
||||
|
||||
<p>If people fear fear itself, they'll deny danger because they don't want to create "mass panic". The problem's not fear, but how we <em>use</em> our fear. Fear, used well, gives you energy to deal with current dangers, and prepare for future dangers.</p>
|
||||
|
||||
<p>Honestly, the two of us (Marcel, epidemiologist + Nicky, artist/coder) are worried about the future. We bet you are, too. That's why we want to channel <em>our</em> worries into making these <strong>playable simulations</strong>, so that you can channel <em>your</em> worries into understanding:</p>
|
||||
|
||||
<ul>
|
||||
<li><strong>The Last Few Months</strong> (epidemiology 101, SEIR model, R & R<sub>0</sub>)</li>
|
||||
<li><strong>The Next Few Months</strong> (lockdowns, contact tracing, masks)</li>
|
||||
<li><strong>The Next Few Years</strong> (vaccines, loss of immunity?)</li>
|
||||
</ul>
|
||||
|
||||
<p>This guide is meant to give you hope <em>and</em> fear. To beat this virus <strong>in a way that also protects our mental & financial health</strong>, we need optimism to create plans, and pessimism to create backup plans. As Gladys Bronwyn Stern once said, <em>“The optimist invents the airplane and the pessimist the parachute.”</em></p>
|
||||
|
||||
<p>So, buckle in: we're about to experience some turbulence.</p>
|
||||
|
||||
<hr>
|
||||
|
||||
<h1 id="toc_2">The Last Few Months</h1>
|
||||
|
||||
<p>Pilots use flight simulators to learn how not to crash planes.</p>
|
||||
|
||||
<p><strong>Epidemiologists use epidemic simulators to learn how not to crash humanity.</strong></p>
|
||||
|
||||
<p>So, let's create a very simple "epidemic flight simulator"! Here, we have some (i) Infectious people & some not-yet-infected (s) Susceptible people. (i)s turn (s)s into more (i)s:</p>
|
||||
|
||||
<p>// pic</p>
|
||||
|
||||
<p>At the start of a COVID-19 outbreak, it's estimated that the virus jumps from an (i) to an (s) every 4 days.<sup id="fnref1"><a href="#fn1" rel="footnote">1</a></sup> (<em>On average.</em> Remember, there's lots of variation.)</p>
|
||||
|
||||
<p>Here's a simulation of a population with <em>just</em> 0.001% (i) and 99.999% (s), over 6 months. If we simulate "double every 4 days" <em>and nothing else</em>, what happens?</p>
|
||||
|
||||
<p><strong>Click "Start" to play the simulation! (Afterwards, you can re-play the simulation with different settings)</strong></p>
|
||||
|
||||
<p>// sim</p>
|
||||
|
||||
<p>This is the <strong>exponential growth curve.</strong> Starts small, then explodes. "Oh it's just a flu" to "Oh right, flus don't create <em>mass graves in rich cities</em>". </p>
|
||||
|
||||
<p>// pic - exponential double rice</p>
|
||||
|
||||
<p>But, this simulation is wrong. Exponential growth, thankfully, can't go on forever. One thing that stops a virus from spreading is if others <em>already</em> have the virus:</p>
|
||||
|
||||
<p>// pic - 100% spread, 50% spread, 0% spread</p>
|
||||
|
||||
<p><strong>The more (i)s there are, the faster (s)s become (i)s, but the fewer (s)s there are, the <em>slower</em> (s)s become (i)s.</strong></p>
|
||||
|
||||
<p>Now, what happens if we simulate that?</p>
|
||||
|
||||
<p>// sim</p>
|
||||
|
||||
<p>This is the "S-shaped" <strong>logistic growth curve.</strong> Starts small, explodes, then slows down again.</p>
|
||||
|
||||
<p>But, this simulation is <em>still</em> wrong. We're missing the fact that (i) Infectious people eventually stop being infectious, either by 1) recovering, 2) "recovering" with lung damage, or 3) dying.</p>
|
||||
|
||||
<p>For simplicity's sake, let's pretend that all (i) Infectious people become (r) Recovered. (r)s can't be infected again, and let's pretend – <em>for now!</em> – that they stay immune for life.</p>
|
||||
|
||||
<p>When you're infected with COVID-19, it's estimated you stay (i) infectious for 12 days.<sup id="fnref2"><a href="#fn2" rel="footnote">2</a></sup> (Again, <em>on average.</em>)</p>
|
||||
|
||||
<p>Here's a simulation that starts with 100% (i). Most people recover after 12 days, then most of the remainder recover after another 12 days, then most of the remainder <em>of that remainder</em> recover after another 12 days, etc:</p>
|
||||
|
||||
<p>// sim</p>
|
||||
|
||||
<p>This is the opposite of exponential growth, the <strong>exponential decay curve</strong>.</p>
|
||||
|
||||
<p>Now, what happens if you combine this with the S-shaped logistic curve of infection?</p>
|
||||
|
||||
<p>// pic</p>
|
||||
|
||||
<p>Let's find out. Here's a simulation of an epidemic <em>with</em> recovery:</p>
|
||||
|
||||
<p>// sim</p>
|
||||
|
||||
<p>And <em>that's</em> where that famous curve comes from! It's not a bell curve, it's not even a "log-normal" curve. It has no name. But you've seen it a zillion times, and beseeched to flatten.</p>
|
||||
|
||||
<p>// pic: 3 rules</p>
|
||||
|
||||
<p>This is the the <strong>SIR Model</strong>, ((s) <strong>S</strong>usceptible → (i) <strong>I</strong>nfectious → (r) <strong>R</strong>ecovered) the second-most important idea in Epidemiology 101.</p>
|
||||
|
||||
<p>Note: The simulations that inform policy are <em>far</em> more sophisticated than this! But the SIR model can still help us understand a lot about COVID-19, even if missing the nuances.</p>
|
||||
|
||||
<p>Actually, let's add one more nuance: before an (s) becomes an (i), they first become an (e) Exposed person, when they're infect<em>ed</em> but not yet infect<em>ious</em> – they have the virus but can't pass it on (yet).</p>
|
||||
|
||||
<p>(This variant is called the <strong>SEIR Model</strong>, where "E" stands for (e) Exposed. Note this <em>isn't</em> the everyday meaning of "exposed", where you might or might not have the virus. In this technical definition, "Exposed" means you definitely have it. Yeah, science terminology is bad.)</p>
|
||||
|
||||
<p>For COVID-19, it's estimated that you're in this "latent period" for around 3 days.<sup id="fnref3"><a href="#fn3" rel="footnote">3</a></sup> What happens if we add that to the simulation?</p>
|
||||
|
||||
<p>// sim</p>
|
||||
|
||||
<p>Not much, actually! The "latent period" only changes <em>when</em> the peak happens, but the <em>height</em> of the peak – and total people infected – remain the same:</p>
|
||||
|
||||
<p>// pics</p>
|
||||
|
||||
<p>Why's that? Because of the <em>first</em>-most important idea in Epidemiology 101:</p>
|
||||
|
||||
<p>// pic - <strong>"R"</strong></p>
|
||||
|
||||
<p>Which is short for "Reproduction Number". It's the <em>average</em> number of people an (i) infects <em>before</em> they recover (or die).</p>
|
||||
|
||||
<p>// R > 1, R = 1, R < 1 pic</p>
|
||||
|
||||
<p><strong>R</strong> changes over the course of an outbreak, as we get more immunity & interventions.</p>
|
||||
|
||||
<p><strong>R<sub>0</sub></strong> (pronounced R-nought) is what R is <em>at the start of an outbreak, before immunity or interventions</em>. R<sub>0</sub> is also called the "basic reproduction number". R<sub>0</sub> more closely reflects the power of the virus itself, but it still changes from place to place. For example, because heat 'kills' coronaviruses, R<sub>0</sub> for COVID-19 is lower in hot places than cold ones. Not low enough to contain it, though.</p>
|
||||
|
||||
<p>(A lot of news outlets – and even academic papers! – confuse R and R<sub>0</sub>. Again, science terminology is bad.)</p>
|
||||
|
||||
<p>The R<sub>0</sub> for the flu<sup id="fnref4"><a href="#fn4" rel="footnote">4</a></sup> is around 1.3. The R<sub>0</sub> estimates for COVID-19 are usually between 2 and 3, maybe as high as 6.<sup id="fnref5"><a href="#fn5" rel="footnote">5</a></sup></p>
|
||||
|
||||
<p>In our simulations, an (i) recovers in 12 days, but infects one new (s) every 4 days. That means, <em>on average</em>, an (i) infects 3 (s)s before they recover. So for our simulations, R<sub>0</sub> is 3.</p>
|
||||
|
||||
<p><strong>Play around with this R<sub>0</sub> calculator, to see how R<sub>0</sub> depends on recovery time & new-infection time:</strong></p>
|
||||
|
||||
<p>// calc</p>
|
||||
|
||||
<p>But remember, the fewer (s)s there are, the <em>slower</em> (s)s become (i)s.
|
||||
R depends not just on R<sub>0</sub>, but also how many people are no longer Susceptible – due to, say, having recovered & gotten natural immunity.</p>
|
||||
|
||||
<p>// calc 2</p>
|
||||
|
||||
<p>When enough people have natural immunity, R < 1, and the virus is contained! This is called <strong>herd immunity</strong>, and while it's <em>terrible</em> policy, (we'll explain why later – it's not for the reason you may think!) it's essential to understanding Epidemiology 101.</p>
|
||||
|
||||
<p>Now, let's play the last simulation again, but showing R<sub>0</sub>, R over time, and the herd immunity threshold:</p>
|
||||
|
||||
<p>// sim</p>
|
||||
|
||||
<p>Note: Total cases (the gray curve) does not stop at herd immunity, but <em>overshoots</em> it! And it does this <em>exactly when</em> current cases (the pink curve) peaks. This happens no matter how you change the settings:</p>
|
||||
|
||||
<p>// pic</p>
|
||||
|
||||
<p>This is because, by definition, when there are more non-(s)s than the herd immunity threshold, you get R < 1. And, by definition, R < 1 means new cases stop growing.</p>
|
||||
|
||||
<p>If there's only one lesson you take away from this whole guide, here it is, in big shiny letters:</p>
|
||||
|
||||
<h1 id="toc_3">R > 1 = bad</h1>
|
||||
|
||||
<h1 id="toc_4">R < 1 = good (R=1, meh)</h1>
|
||||
|
||||
<p><strong>This means: we do NOT need to catch all transmissions, or even nearly all transmissions, to stop COVID-19!</strong></p>
|
||||
|
||||
<p>It's a paradox. COVID-19 is incredibly contagious, yet to contain it, we "only" need to stop 67% of infections. 67%?! If that was a school grade, that's a D+. But if R<sub>0</sub> = 3, cutting that by 67% gives us R = 0.99, which is R < 1, which means the virus is contained!</p>
|
||||
|
||||
<p>(And even if, extreme-worst-case, R<sub>0</sub> = <em>6</em>, you still "only" need to stop 84% of transmissions. That's a B grade.)</p>
|
||||
|
||||
<p>// calculator - custom</p>
|
||||
|
||||
<p><em>Every</em> COVID-19 intervention you've heard of – handwashing, social distancing, lockdowns, self-isolation, contact tracing & quarantining, face masks, even "herd immunity" – they're <em>all</em> doing the same thing:</p>
|
||||
|
||||
<p>Getting R < 1.</p>
|
||||
|
||||
<p>So now, let's use our "epidemic flight simulator" to figure out the next few months! How will we get R < 1 in a way that protects not just our physical health, <strong>but also our mental health, social health, <em>and</em> financial health?</strong></p>
|
||||
|
||||
<p>Brace yourselves for an emergency landing...</p>
|
||||
|
||||
<div class="footnotes">
|
||||
<hr>
|
||||
<ol>
|
||||
|
||||
<li id="fn1">
|
||||
<p>source <a href="#fnref1" rev="footnote">↩</a></p>
|
||||
</li>
|
||||
|
||||
<li id="fn2">
|
||||
<p>source <a href="#fnref2" rev="footnote">↩</a></p>
|
||||
</li>
|
||||
|
||||
<li id="fn3">
|
||||
<p>source <a href="#fnref3" rev="footnote">↩</a></ |