Ask HN: Could Omicron boosters create an overfitting problem? | Hacker News Reader

Ask HN: Could Omicron boosters create an overfitting problem?

16 pointsNicoJuicy15 comments

In ML there is a term: overfitting ( https://en.wikipedia.org/wiki/Overfitting )

There is currently a lot of booster talk and it seems it protects ~ 10 weeks.

I'm fully vaccinated ( Pfizer), if I plan to have the booster now then it's reasonable to assume I will get one again after 2-3 months.

In that scenario:

- Vaccine ( = 2 x )

- Booster 1 ( January)

- Booster 2 ( March)

Then in March, the Omicron booster will be ready so it's reasonable to assume I'll receive one again in June. If Omicron protects ~3 months against reinfection and the peak of infection is end of January in Europe.

In the scenario that there becomes a Omicron variant that is more infective and more dangerous, it's not unlikely that it will spread around March-April pretty fast. Before the next booster in June.

Could the human body have an overfitting problem because I was vaccinated too much against the previous variant ( that protects against Omicron, so is used against Omicron too)? This is in the scenario that I haven't received a Omicron booster and there is no "new variant booster" available.

Or is this not how the body works?

I'm curious about any related resources or research based opinion to this hypothesis ( it is a hypothesis, there are many variables that could change it and that's not what i want to discuss/find answers to. ).

Note: I'm just wondering how the human body works in this scenario and I don't know where i can ask my question. HN is a diversified crowd with people more knowledgeable about this.

I think the term you're looking for is "original antigenic sin".

That seems to be highly correlated, thanks!

Omicron is a weird variant. It has evolved to defeat three different therapeutic monoclonal antibodies.

Antibodies are like the immune systems classifiers, a function that takes in a object and returns a probability that the object is a specific pathogen. If the antibody binds to the pathogen for a long time, then P_pathogen is high, if it quickly unbinds or does not bind at all, P_pathogen is low. Brownian motion is constantly bumping around all the tiny objects in the warm soup of your body, so each antibody function can be evaluated thousands of times per second.

In ML terms Omicron is like one of those adversarial image attacks that tricks a classifier into thinking a gun is a duck.

The interesting question is, was Omicron trained by nature or by man? https://bprice.substack.com/p/lab-leak-20

I'm not looking into conspiracy theories. The situation is what it is.

i think it's similar to a inverse adversarial image attack scenario. Since the danger of COVID is the the immune system overreacting. So it's a gun, potentially misidentified as a nuclear bomb and we want 'Stanislav Petrov' as a fix ( https://en.m.wikipedia.org/wiki/1983_Soviet_nuclear_false_al... )

I'm only interested in the trained & best evolution, while applying transfer learning and without being an agent like in NEAT! :P

https://en.m.wikipedia.org/wiki/Neuroevolution_of_augmenting...

There's genetic evidence that the Omicron variant evolved in mice:

> Scientists at the Chinese Academy of Sciences now report that they have found a pattern of mutations in Omicron that is typical of a virus that has infected mice.

> They believe SARS-CoV-2 jumped from a human to a mouse in mid-2020, then back into a human in late 2021.

> They say it is unlikely that the intermediate host was a laboratory mouse.

[0] https://www.medicalnewstoday.com/articles/covid-19-did-omicr...

Yes but why no silent mutations?

At least 10 silent mutations? https://www.forbes.com/sites/williamhaseltine/2021/12/08/omi...

Hmm so would you say omicron evolved naturally without human intervention?

I don't see why humans would have done this without taking credit for it.

The risk is much lower and what i read, the evolution is normal when it would have evolved through mice.

I'm not an expert, but it seemed plausible when I read it.

https://www.medicalnewstoday.com/articles/covid-19-did-omicr...?

I'm not a believer of a man made virus. A coronavirus would have been a bad choice either way because it mutates too easily.

It's also what I read, it could be the reason why the omicron variant is less dangerous.

It doesn't make sense to do this in a lab setting without taking credit for it ( for making it less dangerous).

But this is still not sure ofc. It's just guessing at this point.

Not an expert but please take a look at https://en.wikipedia.org/wiki/Original_antigenic_sin

Jup, josephcsible mentioned it before and it got me into a rabbit hole that resulted to the following research:

https://www.cell.com/trends/immunology/fulltext/S1471-4906(2...

I'm receiving external feedback based on my findings later on. I have a temporarily conclusion, but it's debatable because of lack of domain knowledge.

Tldr: too early to conclude anything with reasonable certainty

Yes there is so much uncertainty that I think that the best response is to do nothing.

I'm not sure if that's the thing. It seems that "original antigenic sin" was already considered ( as a variable) from the beginning. According to what i read, it could be that mRNA vaccines don't have that problem as much ( or perhaps at all?)

I'm waiting for external feedback though. I'll see what it gives.

I don't see the point of vaccinating against a lesser variant. They should save the powerful mRNA effectiveness for deadlier variants.

Reply on news.ycombinator.com