Published using Google Docs
Intentional Immunity Through Voluntary Exposure
Updated automatically every 5 minutes

Ashok Basawapatna. Intentional Immunity Through Voluntary Exposure. March 22, 2019.

Creative Commons LicenseThis work is licensed under a Creative Commons Attribution 4.0 International License.

Intentional Immunity Through Voluntary Exposure[1] 

You can also find a prettier looking version of this article on Medium by clicking here

Assuming we can effectively predict COVID-19 severity of infection in people: a simulation-based argument for enabling people to opt in to contracting COVID-19 in order to avoid a post-lockdown resurgence, give people their lives back, and ensure the safety of vulnerable populations

By Ashok Basawapatna

Disclaimer: I am not an expert in this field, I am writing this article because I think that Intentional Immunity Through Voluntary Exposure should be seriously considered as a possible solution to the COVID-19 pandemic. Please do not take this action without official direction as it could be dangerous to you and the people around you. This article is purely for the purposes of discussion.

Introduction

What if we had a vaccination for COVID-19 that was effective against up to 80% of the population? Well, we have that, it's called the human body.

Upon getting infected with COVID-19, 80% of the population will experience mild cold-like symptoms, and as recovery begins, will not pose a serious risk of infecting others. Important notes, mild symptoms can still mean some harrowing experiences, and ages 20 to 44 make up 20% of the hospitalized US population from COVID-19 with a 0.13% death rate. Importantly, according to a large study out of Iceland, an estimated 50% of the population may have no symptoms whatsoever. Assuming we can develop mechanisms to identify who will get mild or no symptoms, we can immediately start to give people immunity through exposure. For example, looking at age[2] and other factors we can begin building a profile of groups that might not be at risk and thus candidates for Intentional Immunity. Later in the paper, we show that with just 32% of the population opting into being exposed to COVID-19, we begin to see deep benefits in terms of stifling the spread of the virus, decreasing the total sick at any one point, and decreasing the total percent dead, meaning we can be even more selective to ensure safety. Additionally, after recovery, they will be free of symptoms, free from spreading the virus, and able to go back to work in a matter of weeks (it should be noted that anecdotally some doctors are beginning to dispute this characterization of the virus as they begin to see different populations getting sick, but generally, the 80% number is widely reported).

If these people opt in to being intentionally infected, it will take away the guesswork as to whether a person is infected or not-- they know to stay home quarantined from anyone who is also not intentionally infected.[3]  Furthermore, since it is known that those who opt in have the virus, special medical supervision and reserved resource allocation can be given to this group of people, enabling quick reaction in case something goes wrong. This is in contrast to people with mild symptoms unknowingly incubating the virus and spreading it to vulnerable populations they come into contact with.

In addition to saving lives, If 80% of the population really did opt in, in 2 weeks,[4] around 80% of all our businesses (pure guesstimation) could get back to partially or mostly functioning. Our cities and states, like California which is the 5th largest economy in the world, would no longer completely grind to a halt. We would be saving people jobs, money, and alleviating psychological problems.  It is not unthinkable that people within the demographics for which symptoms are typically mild would opt in to Intentional Infection in exchange for their livelihoods,  freedom, and in-person social contact. Finally, and maybe most importantly, it would decrease the chance of infection for people in the 20% as less people in the general population would be potential carriers.

However, what makes this approach even more important is that in our rush to flatten the COVID-19 infection curve, through quarantine and lockdown, we may actually be making ourselves more vulnerable to COVID-19 down the road. This is because the quarantine does not change the underlying reasons that make COVID-19 so destructive in the first place. Namely, if our quarantine is effective and cases die down, it is possible we still have a majority of the population not immune to the virus, still having mild symptoms, still acting as carriers, and enabling subsequent outbreaks to occur, overrunning hospitals, and hurting the most vulnerable among us. Recently, the W.H.O. even stated that lockdowns by themselves may not be enough to contain a resurgence of the virus and experts across the world are fearing a post-lockdown resurgence in Wuhan.

Instead, we can decrease our susceptibility through a method I call Intentional Immunity Through Voluntary Exposure (IIVE),[5] and it can be accomplished during the city and state lockdowns that are already occurring. Allowing for Intentional Immunity will make these lockdowns even more effective by helping us avoid a resurgence, while allowing those who opt in to infection a chance to get back to normal life.

In this article I outline, through a few simple simulations, how our current lockdown strategy to the COVID-19 virus could end up making us just as vulnerable post-lockdown, and how Intentional Immunity Through Voluntary Infection would improve the results of the lockdown decreasing the chance of subsequent lockdowns. Results of the modeling show that we begin to see huge benefits with only 40% of the 80% (32% of the total population) opting in and present a picture (Fig. 7) of how average death rate and average peak total sickness decrease at different IIVE opt in percentages. Finally, we end with a brief discussion of this strategy including shortcomings and logistic hurdles.

The Simulation

To better explore the idea of Intentional Infection, I built a simulation of how a generic disease spreads. George Box famously said, "essentially all models are wrong, but some are useful." We will make many simplifying assumptions and then run the simulation under a few conditions to see what happens. In a later article, probably a video, I will walk through how to create this simulation and the parameter tradeoffs. If you would like to explore the simulation you can find it here. I won't describe it too deeply in this article, but you can change the Simulation Properties by clicking the gear menu on the top left allowing you to adjust parameters like probability of infection and death. You can find an explanation of the parameter values you can adjust here.  It is written in AgentCubes Online and you can remix it with a login and a license but you can also just reimplement the rules in any other agent-based environment you like.[6]

Our agent-based model consists of 2 important agents. One of them, called “theTwenty," represents the 20% of the population for which infection of this virus is deadly. The other important agent, called “theEighty," represents the 80% of the population for which infection leads to mild symptoms. Each of these Agents has 2 forms-- a normal form and a sick form which is represented by changing their color to green.

 

Fig. 1: Notable Agents in the simulation. theEighty represents the 80% of the population which will suffer mild symptoms from COVID-19 and theTwenty represents the 20% of the population which will suffer severe symptoms from COVID-19

The simulation progresses like a board game with discrete rounds. Each round theEighty and theTwenty agent instances do the following:

1) Move

2) Get sick (with a given percentage-- think of this as a die roll-- if they are healthy and next to another sick agent)

3) Recover (with a given percentage if they are sick)

4) Die (with a given percentage if they are sick)

As one might expect there are some differences between these two agents which can be adjusted in the Simulation Properties. Simply, theEighty moves around a lot more than theTwenty, and when theEighty get sick they have a lower chance of death than theTwenty. For either sick agent, until they recover or die, they can move around and infect healthy people of either group who happen to be next to them.

On the other hand, theTwenty are not as mobile, and becoming infected kills them with a probability at around 14% in two simulation rounds (theEighty dies with a probability of 1%).[7]

The simulation level is set up as follows:

Fig. 2: The Simulation setup: 500 people: 100 instances of theTwenty agent plus 400 instances of  theEighty agent

As shown in Fig. 2, there are 500 agents in total in the level. Clustered in groups of 25 in the four corners are 100 instances of theTwenty agent. Clustered in the middle are 400 instances of theEighty agent. There is space in between theTwenty and the theEighty agents ensuring the virus cannot immediately jump from theEighty to theTwenty. Before the simulation begins, around 3% of theEighty population will be randomly infected with the virus.[8] At this point, the simulation can begin.[9] 

Case 1: No quarantine

With no quarantine, the agents will move around randomly infecting each other and spreading the sickness. In this simulation the pandemic ending is represented by every sick agent instance dying or recovering. In the recovery case, their immunity is set such that they cannot get sick again. Click here to watch a video of no quarantine simulation running.  

Fig. 3 is the plot produced by the simulation in the video with percent dead in red and percent sick in green. The x-axis is simulation time.

Fig. 3: Plot of no quarantine case: y-axis is percent of total population, x-axis is simulation time. It ends with 9.8 percent of the population dead.

At the peak, around 40% of the population has this sickness and it ends with around 9.8% of the total population dead. On average, running this simulation ten times, yields an average of 9.76% dead.[10] Obviously this is much more deadly than COVID-19 in which, even in the worst case estimates, will only kill 0.5 percent of the population. This 9.76% dead creates a baseline for what this fictional simulation virus would do without any intervention.

Case 2: Perfect quarantine

In the perfect quarantine case, the same 3% of theEighty agents still start off sick, however everyone is perfectly quarantined from one another, meaning no interaction between anyone. It is a pretty boring simulation so I'll spare you the video.[11] In the Fig. 4 below, 0.2% of the population ends up dying which is 1 person. Running the simulation 10 times and averaging the percent dead results in 0.14% of the population dying. This is not too surprising and not too realistic. If we perfectly quarantine everyone, the people who initially have the virus will die or recover, without spreading the virus, and the pandemic will be over. Since the only people initially infected are part of theEighty, the chance of them dying are minimal.

Fig. 4: Plot of perfect quarantine case: y-axis is percent of total population, x-axis is simulation time. Green line represents percent sick and red line represents percent dead.

This illustrates an important point. Someone looking at this might think-- "hey, quarantines work great!"-- we went from an average of 9.73% dead to 0.14% dead. Looking at a Value of Statistical Life estimate[12] and assuming each death costs the US $5 million dollars we, in this simulation world with this deadly disease, are potentially saving $240 million dollars, per 500 people, by employing a perfect quarantine. Even if the quarantine is not perfect, as one might intuit, we would still save a ton of money and have the advantage of not creating a run on the hospitals. From all this one might conclude that city lockdowns are good, and we are actually saving money in the long run.

However, thinking a bit more deeply about the simulation, once the perfect quarantine period is finished, the quarantine will be lifted. And since only a few people got the sickness, few in the population are immune. This situation is analogous to a dangerously dense and dry forest where a fire starts but is immediately contained, maybe by a swift summer rainstorm, with only a few trees burned. The containment of that particular fire still makes the dense and dry forest susceptible to an ember. In our cities and states we are not a closed system, and thus, all it takes is one traveler for another viral outbreak if enough of us do not have immunity. At the end of the quarantine, we have survived the virus without changing the underlying conditions of what in part makes this virus so deadly to our population in the first place: the 80%, unbeknownst to them, infecting the other 20%.

Case 3: The Status Quo. Modified Quarantine Case where the quarantine is lifted when the percentage of people sick falls below a given threshold

To further clarify the danger we face post quarantine, we can do a modified quarantine wherein the quarantine is lifted when the sick population gets below a certain percent (in this example, less than 1% of the population infected). So we'll start at around 3 percent, the amount of cases will begin to rise, a quarantine will be automatically instituted when the percent sick gets above 4%. Note, the number of sick cases can jump in one round so it can be any number above 4% before the quarantine is instituted. Hopefully, the quarantine will ensure that the curve will flatten. Finally, when the cases get below 1%, the quarantine will be lifted. This is what I assume to be akin to what might happen in a city like New York and seems to be happening right now in parts of China; at some point we will say we have "flattened the curve" sufficiently and things can begin to go back to normal.[13] 

Click here to see a video of the Modified Quarantine case simulation run. 

Fig. 5 is the plot that results from that simulation run:

Fig. 5: Plot of modified quarantine case: y-axis is percent of total population, x-axis is simulation time. Green line represents percent sick and red line represents percent dead.

The resulting plot is horrifying, especially so to anyone who is living paycheck to paycheck and/or just lost their job due to this lockdown. We initially jump to 15% infected at the beginning of the simulation, and a quarantine is automatically instituted. The quarantine flattens the curve and gets the cases back down to under 1%. So far so good. At this point the quarantine is lifted and the number of sick people shoots back up, and we are put in the same exact situation we were in before the quarantine. In this case, at its maximum point, the percent sick ends up actually being higher than it was at the point the quarantine was instituted.

This turns out to be a particularly good outcome for this scenario as only 7% died, which is an improvement over the no quarantine case.[14] However, running ten trials of the modified quarantine case leads to an average percent dead of 9.6%. This is only slightly better than the 9.76% death rate in the no quarantine condition, yielding a savings of around $4 million dollars in simulation money. This definitely is not close to the $240 million dollars in simulation money we save in the perfect quarantine condition. Though this is an improvement, it is only a slight one, and I imagine many might wonder why we are upending people's lives for this potentially ineffective strategy. Furthermore, the simulation assumes a lockdown or quarantine is only instituted once; more realistically however, if the number of sick people ramped back up, another quarantine would probably be instituted to support the overrun hospitals. Again note that in the case depicted above, the percent sick gets to its worse point after the quarantine. The indication is that post-quarantine, our hospitals can still be overrun. This leads us to our final case which is the proposed alternative to this status quo.

Case 4: Intentional Immunity Through Voluntary Exposure

To model Intentional Immunity Through Voluntary Exposure, we will give a percentage of the population immunity at the start-- these will be the ones that opt in to being infected. Just like the previous simulations, when the percent sick gets above a certain threshold, a quarantine will be instituted. One can think of the Voluntary Exposure of the virus occurring during the quarantine.[15] Though some start off with immunity, the simulation ensures that the same number of the non-intentionally infected start off with the virus as was in the previous cases.[16] Case 3, representing the status quo, can also be thought of as Case 4 with zero percent of theEighty opting into being voluntarily infected.

Here is a sample simulation run video assuming 60% (of theEighty, so 48% of the total population) opt in to being infected and enduring the mild symptoms.

Fig. 6 is the plot from that 60% opt in simulation run

Fig. 6: Plot of the intentional immunity case with 60% opt in: y-axis is percent of total population, x-axis is simulation time. Green line represents percent sick and red line represents percent dead.

Just like the Modified Quarantine Case 3, the number of infected shoots up to 8% of the population, a quarantine is instituted, the number of cases decline. When the percent sick gets under 1% of the total population, the quarantine is lifted and deaths begin to increase. However, at this point we hope that the decreased density in susceptible people keeps the pandemic from exploding after the quarantine is lifted. It looks like, in this particular case, that is exactly what happens, and we end with just 2.4 percent dead.[17] In reality, we might strategically hope that people who come into contact with a lot of other people volunteer to be infected.[18] 

In order to get a better idea of how many people need to opt in to Intentional Immunity for it to be effective, we need to run the simulation multiple times for each opt in case. Running the simulation ten times for multiple opt in percentages yields the following table.[19]

Table 1:  % opting into Intentional Immunity with Average % death and Average % sick at peak

% of theEighty that opt in to Intentional Immunity

Avg % death of total population

Avg % sick at peak of total population

0

9.6

20.96

10

9.26

18.42

20

7.04

13.86

30

6.88

11.4

40

3.82

8.11

50

2.48

5.5

60

2.02

4.4

70

1.1

3.48

80

0.5

2.16

Fig. 7 is a plot of the above table data.

Fig. 7: Plot of Average percent death and Average percent sick at peak vs. percent of theEighty that opt in to Intentional Immunity

As more people of the theEighty opt in to Intentional Immunity, the death rate seems to decrease and the peak percent of people sick at any one time decreases. This implies less deaths and less overtaxing of hospital resources. Benefits occur even when a majority of theEighty do not opt in to Intentional Immunity Through Voluntary Exposure. For example, at 40% of theEighty opting in, which is 32% of the full population, cuts the percent death from 9.6% of the population to 3.82% saving $144 million dollars of simulation money over the status quo (Case 3) wherein 0% opt in. At 80% opting in, 64% of the full population, a majority of the full population is inoculated dropping the average death to 0.5% and saving $227 million dollars in simulation money over the status quo (and $231 million over the Case 1: the no quarantine case). These results seem to highly indicate that Intentional Immunity Through Voluntary Exposure deserves to be seriously considered and discussed as a possible way to combat the COVID-19 epidemic.

Discussion  

COVID-19 is a particularly destructive virus because of its ability to be relatively harmless to 80% of the population. Intentional Immunity through Voluntary Exposure exploits this idea to inoculate a portion of that 80% of the population. This can occur during the current lockdown and decrease the chances of another "2nd wave" COVID-19 outbreak when cities and states all across America begin to lift their restrictions. Furthermore, by keeping track of who voluntarily opted in, we can allow them to return back to their pre-COVID-19 life after enough time has elapsed such that they are no longer contagious (as mentioned above, around 14 days). In effect, one or two COVID-19 tests can act as a vaccination for the 80% of the population that will experience mild symptoms. We can further decrease the risk to participants who opt in by only taking a subsection of this population who are the candidates most likely to not have any complications. This opt in benefits the vulnerable 20% in that the people who opt in are no longer spreading the virus yielding a lower density of susceptible carriers in the community.

The reasons Intentional Immunity Through Voluntary Exposure might not be effective include the idea, as mentioned in the introduction, that people we initially believed might be in the 80% of the population are actually reacting worse to contracting COVID-19. If this is true, then we have to find a better way of identifying who would react mildly to COVID-19 and who would not before instituting Intentional Immunity. Looking at the chart in this article, it seems pretty clear that we can safely identify groups that are low risk and make them even lower with medical supervision upon opt in.

Furthermore, since tests are in scarce supply, testing someone to ensure they have actually contracted COVID-19 and then retesting them after to ensure they are cured might not be possible until more tests are procured. However, some are estimating lockdowns to continue for 4 or even 8 weeks (or 4 to 9 months!) meaning as more tests come in, there will be a larger opportunity to implement Intentional Immunity.

This is just an idea and much of the logistics have to be worked out. For example, details on how this is implemented including how we screen for people who are opting in, how and where we test them before and after they get the disease, and how we allow them to have the necessary permissions to go back to work and hangout in public during a lockdown are all valid questions that this article does not answer. Moreover, there is an assumption that it is easy to infect people, but the logistics of infecting people intentionally may prove more challenging. Finally, no one likes to be sick but this paper makes the assumption that people would willingly trade a few days of mild sickness (which, again, can still be uncomfortable or even worse, intense) to go back to a normal life including human social contact with others. Despite all these logistical hurdles, it is in our best interest to figure out a way we might implement this for COVID-19 and subsequent viruses that have these characteristics. Figuring out how to overcome these challenges will enable use of Intentional Immunity Through Voluntary Exposure for any virus that affects a large portion of the population mildly but is deadly to others-- which seems to be the calling card of pandemics.

Currently a quarter of the US economy is on lockdown. Unemployment is soaring. The benefit of allowing workers who survive COVID-19 to go back to work can help in part stem the imminent economic disaster. An estimated 75% of Americans live paycheck to paycheck. Allowing people to go back to work will free up money for others who more rely on government assistance to get through this time in lockdown. 

COVID-19 vaccines are currently being developed, and they must be developed to protect the members of the vulnerable 20%. IIVE is already being explored as a method to research vaccination development. The estimates for vaccine development state that it could be 12 to 18 months before a vaccine can be provided to the public. The first article above explains that 86,025 people have recovered from COVID-19 at publication time, and their blood can potentially be processed for antibodies. Intentional Immunity Through Voluntary Exposure supports this strategy by increasing the pool of those who can donate life saving antibodies decreasing the burden on any one person.

When people are locked inside and freedoms curtailed, it is time to think more critically, not less, about whether these actions will produce the desired outcome. What seems most surprising to me and many others is that there has not been a discussion with the American people as to why there are no alternatives to locking everyone down for something that has minor effects on 80% of the population. The lockdown strategy seems like a hastily developed drastic measure in response to a virus we were woefully unprepared for. At this moment, we can take a deep breath, see if there are not any better alternatives, and begin to be more intentional in our strategy combating this virus as opposed to locking everyone up and merely hoping that it does not spread or that hot weather will kill it.  I believe Intentional Immunity Through Voluntary Exposure might be one of these better alternatives, and if you agree, or have other alternatives that need to be discussed, please share them.

Ashok Basawapatna is an Assistant Professor of Math/CIS at SUNY Old Westbury and is staving off boredom hiding out at his parents' house in Denver writing google docs and preparing teach online classes. He thanks his mom, dad, and brother for their article feedback and food. Furthermore, his brother says he has no interest in opting in.

Ashok Basawapatna. Intentional Immunity Through Voluntary Exposure. March 22, 2019.

Creative Commons LicenseThis work is licensed under a Creative Commons Attribution 4.0 International License.


[1] This work is licensed under a Creative Commons Attribution 4.0 International License.

[2] The linked article states 0.5% fatality in ages 15 to 44 and 0% fatality in younger kids. However, the additional argument is that we can give those who voluntarily opt in special medical supervision further decreasing the already small percentage of fatalities among those populations.

[3] Hypothetically, intentionally infected people could hang out together

[4] The Report of the WHO-China Joint Mission on Coronavirus Disease 2019 (COVID-19) states, "Using available preliminary data, the median time from onset to clinical recovery for mild cases is approximately 2 weeks and is 3-6 weeks for patients with severe or critical disease."

[5]  Many have thought of this concept already and thus, there might be different names for it. Some parents reading this may have even intentionally exposed their children to chicken pox.

[6]  AgentCubes online was  created by AgentSheets, Inc. with SBIR grants from the NSF. Full disclosure: My thesis advisor, Dr. Alexander Repenning, worked on its creation, and it is a part of my research related to Computer Science Education

[7] For multiple reasons it is hard to recreate the parameters of the COVID-19 virus in a simple simulation. Therefore, think of this as a virus with similar properties. Specifically, it is being carried by 80% of the population, of whom it is not deadly for, but becomes a serious illness when contracted by 20% of the population.

[8] In practice 3% was typically not reached, the initial infection percentage typically hoovered around 1-2%

[9] Note this simulation is clearly not a perfect representation and there are a lot of implementation details I am not going into. Some of the choices, like densely packing the theEighty together and sequestering theTwenty on the sides of the level might make sense or not-- feel free to redo the simulation, in a way you find more authentic, to see the results.

[10] you can check out all the raw simulation data reported in this article here

[11] You can recreate it by setting the Quarantine Simulation Property to 1 and the AutoQuarantine property to 0

[12] These estimates make me uncomfortable but I have to believe, and that article seems to give credence to the idea, that this estimation was used to justify instituting lockdowns. Therefore, we will use them here.

[13] If you are playing along at home, you can accomplish this by setting Autoquarantine, Mqlimitpercent, and Modifiedquarantine to 1 in the Simulation Properties and Autoquarantinepercent to 4 and Quarantine to 0.

[14] Just like a boardgame, a single simulation run can just be terribly lucky or unlucky because of the chance involved

[15] This is implemented slightly differently in the simulation, agents are given immunity at the beginning but since quarantine happens almost immediately in all cases it makes little difference

[16] This is done automatically by increasing the initialSickPercent when the simulation starts based on the increase of Intentionalinfectionpercent in the Simulation Properties.

[17] It should be noted that we can again be lucky or unlucky with a simulation run and in real life. For example, a group of people who opt into not being infected all clustering in one region means that the rest of the possibly infected population is directly connected yielding an scenario where if one of them gets sick the disease has a good chance to spread.

[18] Network Theory provides one way these strategically important people can be identified. 

[19] Again, raw data can be found here