HELP Please 🙏 We're developing a CO2 calculator for the AirSpot App. Has anyone got access to or able to develop an equation (based on Hadrell's data) where a CO2 level can then be turned into an estimate of COVID infection risk? #CleanIndoorAir #COVIDisAirborne www.nature.com/articles/s41... - ThreadSky

airspotrob.bsky.social • 200 days ago

HELP Please 🙏

We're developing a CO2 calculator for the AirSpot App.

Has anyone got access to or able to develop an equation (based on Hadrell's data) where a CO2 level can then be turned into an estimate of COVID infection risk?

#CleanIndoorAir #COVIDisAirborne

https://www.nature.com/articles/s41467-024-47777-5

Comments

suprion.bsky.social•198 days ago

📌

suprion.bsky.social•198 days ago

Did you contact him?

You may want to drop me an email …

fusionprogguy.bsky.social•200 days ago

Try this - based on some of the source data in the first paper you shared.

Infectivity = 0.443 + 0.000166*CO2 + (241*RH^2 - 0.0227*CO2*RH)/(179 + seconds) - 0.622*RH - 1.06131543614508e-8*CO2*seconds - 1.06131543614508e-8*CO2^2

Happy testing!

fusionprogguy.bsky.social•200 days ago

Here are some heat maps with C02 vs RH, CO2 vs Time, RH vs Time

fusionprogguy.bsky.social•200 days ago

Infectivity vs C02, Infectivity vs RH, Infectivity vs Time
#covidisairborne #C02 #humidity #ventilation #airquality

fusionprogguy.bsky.social•199 days ago

If you don't want to use the "seconds" variable and just CO2 and relative humidity (RH), you can use

Infectivity = 0.815 - 0.115*sqrt(sqrt(0.162/RH + CO2*1.38e-9^(1.34e-9*RH*CO2^3)))

fusionprogguy.bsky.social•199 days ago

I applied some data smoothing and came up with this Infectivity formula with C02 and Relative Humidity

Infectivity = 0.04073*RH + 0.000221*CO2 + 7.35e-6*RH^3 - 0.207 - 2.6e-8*CO2^2 - 0.000997*RH^2

fusionprogguy.bsky.social•199 days ago

This is what the smoothed C02 data looks like with the formula.

ukhadds.bsky.social•196 days ago

What data is used to make these figures? And does infectivity=transmission, or virus viability?

fusionprogguy.bsky.social•193 days ago

It’s from the nature paper that was shared in the original post. You can download the supplementary csv file with the data collected in the study.

ukhadds.bsky.social•192 days ago

In that paper the airborne survival is always higher at higher RH. For SARS-CoV-2, there isn't the characteristic U-shape associated seen for influenza.

hallinen.bsky.social•196 days ago

I recall, perhaps incorrectly, that the sweet spot for SARS2 demise is around 50% humidity. Maybe at 90% virus containing respiratory aerosols are sticking to mold spores. I love the idea of a calculator.

ukhadds.bsky.social•195 days ago

The relationship between humidity and infectivity is clear (survives a bit better at higher humidity), but for transmission the relationship isn't very clear. I put together a couple of videos digging into this. I like the idea of a model, I'm just curious what it's based upon. https://youtu.be/WUGUA_1H2Kk

airspotrob.bsky.social•200 days ago

Also this article provides an estimate of infectivity based upon CO2 levels: https://www.sciencedaily.com/releases/2021/04/210407143809.htm

airspotrob.bsky.social•200 days ago

'If CO2 levels in a gym drop from 2,800 to 1,000 ppm (~2,400 above background levels to 600), the risk of transmission drops, too, to one-quarter of the original risk. In the library, if an influx of people makes CO2 jump from 800 to 1,600 (400 to 1,200 above background), transmission risk triples.'

minxmarx.bsky.social•195 days ago

Not an air quality expert but I think the challenge cd be determining the relationship between CO2 & particles in the air (PM2.5). From personal experience--CO2 can be high but PM2.5 and AQI excellent so if this were a public space risk shd be low? (I live alone so no other inputs in room.)

cleanairstars.com•199 days ago

Unfortunately you won’t be able to calculate risk of transmission as its function is dependent on variables you have no ability to assess (emission rate, distance to emitter, airflow direction, etc). Best you can do is calculate relative risk against another scenario with adjusted variables.

airspotrob.bsky.social•199 days ago

Thank you

How do we calculate relative risk from the CO2?

ocnj.bsky.social•199 days ago

Maybe the papers behind these have some math?

alvelda.bsky.social•190 days ago

It’s possible to generate rough estimate curves in the left hand graph that would depend on:

A. The overall prevalence, i.e. in the U.S. right now 1 in 180 people are infectious. And

B. How many people are in the room,

C. Their mask status & quality

D. Air flow & filtration estimates.

alvelda.bsky.social•190 days ago

But really, all you need is the color code on the left hand graph, where the CO2 level is an overall proxy for risk level, as it represents how much exhausted is accumulating.

If you want to keep your COVID risk below, say, 5% for any environment, mask up if CO2 is over 800ppm. If over 2,000, N95!

alvelda.bsky.social•190 days ago

That should have read “exhaled C02” not “exhausted.”

cleanairstars.com•199 days ago

This specifies the conditions under which that risk was calculated (even then it would be challenging to read this chart reliably due to the additional variables of emission rate which can vary by orders of magnitude due to variables such as emitter’s viral load, vocalisation etc)

ocnj.bsky.social•199 days ago

For sure. I don't see how you could come up with a one size fits all formula, and the more variables you add the less it will be used. Just getting people to equate 700 ppm with GOOD will be a battle.

cleanairstars.com•199 days ago

Yes. A high CO2 can tell you there’s room for improvement with ventilation (engineering perspective i.e. modifiable environment factors) or reducing occupancy (public health perspective i.e. numbers of potential emitters and susceptibles that could continue chains of transmission).

cleanairstars.com•199 days ago

There would be at least a couple of considerations.
1. Relative risk from poor ventilation (Need to estimate ventilation which is not straightforward)
2. Relative risk from modified viability (still too early to verify significance)
Goal is deriving some semblance of actionable data.

breathewell.bsky.social•196 days ago

Kaiterra Sense Edge has that calculation. May can derive helpful info from that.

airspotrob.bsky.social•196 days ago

Thank you - I've taken a look at the website - looks great.

Would you be able to provide the link to page discussing infectivity risk? 🙏

breathewell.bsky.social•196 days ago

https://support.kaiterra.com/reset-viral-index

upswept.bsky.social•190 days ago

📌

chronicallthestuff.bsky.social•190 days ago

Like @cleanairstars.com said, too many unknowable values to calculate precise risk with other people around, but I'd be very interested in calculating time to 99% decay from continual CO2 levels when entering a space w/ no other occupants. @ukhadds.bsky.social, are there enough data points for this?

ukhadds.bsky.social•190 days ago

Regarding the effect of a range of CO2 affecting viral decay, I've seen models made with less data.

cleanairstars.com•190 days ago

Ultimately the question is how actionable the output is. What’s the sensitivity and specificity of the number in detecting high risk of infection scenarios. There will be scenarios where CO2 is low and risk of infection is high, and those where CO2 is high but risk of infection is low.

suprion.bsky.social•188 days ago

My impression is that some people here try to establish so called bijection (from maths), where there is none.

To see, assume:
🔸two identical rooms, two identical groups of people
🔸100% infected in group 1, 0% in group 2
🔸both have same CO2(t)
🔸you can't map back from CO2 levels ("risk")

suprion.bsky.social•188 days ago

The "magical" extra parameters some seem trying to find are called Noise in the field of optimizing products and technology.

Upper diagram: from users perspective
As MANUFACTURER ... those "controls" are your additional noises ...

Below: noise in Risk.
Optimization process: N2 turned to M-axis (!)

suprion.bsky.social•188 days ago

Hm, the bijection map got lost above. So here it is again:
🔸mapping set A onto set B
🔸injection: all A find one partner in B, but some B's may go unmapped (one-to-one)
🔸surjection: all A find a partner in B, but B is so small, that some B's are "hit" several times
🔸bijection: BOTH (logical AND)

ukhadds.bsky.social•188 days ago

I get scenarios where the CO2 can be high and the risk is low (eg. very good filtration). Out of curiousity, what kind of situation do you think would be the reverse (low CO2, high risk)?

sourcecontrol.bsky.social•188 days ago

short range in a crowd even outside

ukhadds.bsky.social•188 days ago

Short range? The CO2 is high in the plume. Or do you mean beyond the plume?

cleanairstars.com•188 days ago

In a large space where someone is in close proximity to an emitter and there’s little active or passive mixing. CO2 will generally diffuse more evenly throughout a space than aerosols.

cleanairstars.com•188 days ago

But importantly, it takes longer for CO2 to reach its steady state in a large space. Someone could be in a poorly ventilated space for a while before CO2 reveals poor ventilation.

chronicallthestuff.bsky.social•190 days ago

Cool, anything public access?

ukhadds.bsky.social•188 days ago

https://www.dhs.gov/archive/science-and-technology/estimated-surface-decay-sars-cov-2

chronicallthestuff.bsky.social•188 days ago

Ah right, thanks, I've used that tool to extensively. It seems quite conservative? I was more wondering if there were any similar calculators based on CO2.

ukhadds.bsky.social•188 days ago

Not that I've seen. My larger point was simply that there's more data on the relationship between CO2 and decay in our one paper than all the studies used to make this model.

Comments

Posting Rules

Reply