Research Into Stochastic Systems with AI – Pandemic Relief

By James Carlson

We have all experienced the limited focus of the Center for Disease Control (CDC) under the former head, Dr. Fauci, and the results of the mismanagement of that office. The problem we faced then goes deeper than a person (I’m not anti-fauci but pro-science). The science is not analytic (deterministic) but stochastic and as such many comments and solutions were presented as analytic (supposed to 100% verified) when that is an unreasonable goal. In short, we need to increase research into various stochastic systems using AI to present alternate and viable solutions to treating pandemics.

Getting back to basics, let’s define terms.

·       System – Characterized by 5 things

o   Parameters

o   Interdependent relationships between parameters

o   Initial Conditions

o   Inputs & Outputs (this defines a closed system as an open system)

·       Analytic System

o   Any given system where all the characteristics are known (examples are simple machines)

·       Stochastic System

o   Any given system where all the characteristics are not known (examples are meteorology, economics, and the medical arts)

There are many systems that we know everything about but the medical arts is not one of them. We don’t even know how a simple cell works. As such, we have approximate solutions in terms of percentages that allow us to understand and handle our medical needs. If we did know everything about a medical issue, then it would be analytic. As such, all medical advice and treatments are approximate with a high chance (probability) of succeeding. We leave the best guesses to the pros and not average people like me.

During the last pandemic, we had the whole course of ‘faucis’ telling us with absolute certainty exactly what they didn’t know. They gave us analytical solutions to inherently stochastic medical issues. As such they proved their incompetence in the matters of health that concern us all. We have to get past the ‘white robe’ fiction of the medical professionals and filter their advice with modern technology. Looking behind the curtain at who is pulling the strings, and what strings are being pulled is in order.

For instance, if you have a 33% probability protection against any 2-way transmission of a communicable infection, that roughly describes the protection we have when wearing masks, practicing social distancing, and washing our hands. However, when added together, these probability percentages of protection increase from 33% to 55% for 2 layers of protection and 70% for 3 layers of protection. The percentages are never 0% or 100%. The extremes are false analytical estimates and not real-world information. There is value in using these techniques but they are not 100% effective and should be left up to the individual who uses them.

Now, we have the exaggerated use of vaccines that do not provide 100% protection against an infection especially considering that vaccines require you get infected with a weakened strain of the same infection you want protection against. Once injected into your body, the vaccine gets makes you sick and your body produces anti-bodies thus providing a level of protection against the stronger strain of the same infection. Getting sick is a part of the vaccine process. Most people don’t seem to know this.

Getting sick is also a part of life in general. People, apart from vaccines, who get sick have antibodies produced by their bodies in response to the full-strength of the infection they encountered. And then the antibodies are available for present and future infections. The natural process here is what the artificial process of vaccines imitates. So natural immunity (antibodies) is available to any population of people during an emerging and ongoing pandemic.

These antibodies can be harvested by blood donations from donors who have acquired a natural immunity from the natural process and/or from vaccinated persons who have the same from the artificial process. This natural (antibody) immunity can be gleaned from whole blood into plasma (or serum) that can be injected or transfused into people who have never received a vaccine or been sick with the infection of concern. Thus, we have a protective layer of immunity available to many people in our population that doesn’t require getting sick or getting a vaccine. This is where improvements can be made for any new pandemic and where AI may assist in how that may be applied.

Working with ChatGPT, I posed the question:

o   blood, plasma, and serum transfusions can carry b-cells and antibodies useful for protection against a pandemic. What percentage of coverage for a population would these transfusions cover.

The immediate answer was:

o   Yes, convalescent blood products could provide some temporary population protection in a new pandemic, but the limiting factor would be supply rather than biology. The most useful component would be antibody-rich plasma or purified immunoglobulin, not whole blood.

o   One important correction: plasma and serum transfer antibodies, not meaningful numbers of B cells. Whole blood/cellular products can contain donor lymphocytes, including B cells, but modern blood components are commonly leukoreduced; leukoreduction can remove >99.9% of leukocytes. Those donor B cells ordinarily do not establish a durable immune system in the recipient. Passive antibodies, by contrast, can provide protection within hours, but typically last weeks to months, rather than producing permanent immunity.

So, while I got a positive return for antibodies, I got a negative return on B-Cell transfusions via whole blood.  This first thread from AI concluded:

o   The especially interesting scenario is the first few months before a vaccine exists. Convalescent antibody collection could create a temporary immune "shield" concentrated on healthcare workers, first responders, elderly people, and other high-risk groups. Convalescent plasma has in fact been investigated for both treatment and post-exposure prophylaxis; the principle is passive antibody transfer.

AI then asked:

o   If you're thinking about this as a pandemic-defense system, rather than normal transfusion medicine, we can model something more interesting: how many immune survivors would be needed to passively protect 10%, 25%, 50%, or 100% of a population, including repeat plasma donations and antibody half-life. That would give us an actual coverage curve.

I responded with:

o   yes, continue

To which AI gave a lengthy response as follows:

……….

Yes. We can turn the idea into a simple pandemic passive-immunity coverage model. The numbers below are illustrative rather than a recommendation for transfusion practice; the required antibody dose would depend strongly on the pathogen, neutralizing-antibody titer, route of administration, and clinical evidence.

Define:

where C is the fraction of the total population supplied with one protective course per month, R is the fraction of the population recovered/immune, E is the fraction of recovered people who are eligible and have sufficiently useful antibodies, D is the fraction of those eligible people who actually donate, F is donations per donor per month, and Y is effective recipient courses obtained from each donation.

For a middle-of-the-road emergency scenario, suppose 50% of recovered people qualify, 20% of them participate, they donate 4 times per month, and each donation ultimately yields 1 effective recipient course. Four donations per month is within the U.S. regulatory ceiling for source-plasma collection; FDA materials state that plasmapheresis may occur no more often than every 48 hours and twice in seven days.

That gives:

So the monthly production capacity would look approximately like this:

Population already recovered	Courses produced/month	Population potentially covered
5%	2% of population	2%
10%	4%	4%
20%	8%	8%
30%	12%	12%
50%	20%	20%
70%	28%	28%

This means that once 20% of a population had recovered, a moderately aggressive plasma program might theoretically generate antibody courses for roughly 8% of the population each month under these assumptions.

The situation improves considerably if a donation contains enough high-titer antibody for two useful recipient courses. COVID studies illustrate why this is plausible as an order-of-magnitude assumption, although not a universal rule: trials used doses ranging roughly from 200–500 mL, sometimes divided into multiple transfusions. If processing or unusually high antibody titers permitted two effective courses per collection, the equation becomes:

Then 20% recovered → about 16% of the population supplied per month, 30% recovered → 24%, and 50% recovered → 40%. What would it take to protect 10%, 25%, or 50%?

Using that more productive scenario:

Desired simultaneous coverage	Recovered population required
10%	12.5% recovered
25%	31.3% recovered
50%	62.5% recovered
75%	93.8% recovered
100%	125% — impossible in one month

So universal protection through direct plasma alone is unlikely. But protecting a strategically selected 10–25% of the population is much more conceivable.

There is another important factor: protection need not necessarily be renewed every month. CDC describes passive immunity as immediate but generally lasting weeks to months. If an antibody preparation protected a recipient for, say, three months rather than one month, production accumulating over three months could theoretically maintain a much larger protected pool.

For example, in our scenario with 20% recovered:

So, ignoring repeat dosing, losses, overlapping recipients, antibody decay and continuing infections, there could theoretically be enough antibody courses over three months for 48% of the population.

That does not mean 48% would necessarily be clinically protected. It means the donor-production arithmetic is no longer obviously prohibitive.

The more promising version

I think the strongest version of your idea is therefore not mass whole-blood transfusion. It would be:

recovered population → frequent plasma collection → antibody screening → select high-titer plasma → pool/purify antibodies → hyperimmune immunoglobulin → targeted prophylaxis

FDA already has regulatory guidance specifically addressing collection of source plasma containing disease-associated IgG antibodies. Industrial fractionation can turn antibody-rich donor plasma into standardized hyperimmune immunoglobulin preparations, with pathogen-reduction and purification steps.

That approach also avoids trying to transfer B cells. The useful transferable resource is the antibody repertoire produced by the recovered population.

The pandemic strategy becomes particularly interesting if allocation is prioritized. You would not initially need to cover 100% of society. If the high-risk/critical group were, for example, 15% of the population, then under our illustrative model, once roughly 19% of the population had recovered, the donor pool could theoretically generate enough antibody courses each month to cover that entire 15% priority group:

So your concept could potentially function as a bridge-immunity system: survivors from the first epidemic wave supply temporary antibody protection to healthcare workers, vulnerable people, essential personnel and uninfected high-risk individuals while vaccines and pathogen-specific drugs are being developed.

The next useful step would be to put this into an epidemic-time model—for example, a population of 1 million in which infections rise from 1% → 5% → 10% → 20% recovered, while recovered donors progressively generate antibody protection. That would tell us whether the protective donor pool can grow faster than the epidemic itself, which is the critical test of the concept.

……….

Shifting gears from vaccines only, to include accelerated blood donations from immune donors during a pandemic, will provide needed relief to many people. It remains for the health care professionals to implement these methods, which may be sufficient to covering at least 10% to nearly 30% of a population during a pandemic. Dealing with stochastic systems using AI has great promise. It remains to be seen if these ideas will become reality. Go forth and conquer.