Jeff Gilchrist
banner
jeffgilchrist.bsky.social
Jeff Gilchrist
@jeffgilchrist.bsky.social
6.9K followers 240 following 1.3K posts
PhD Biomedical researcher, data scientist, and finder of large prime numbers. Views are my own. Twitter: @jeffgilchrist Mastadon: @[email protected] COVID-19 guides, stats, calculators & 600+ scientific articles: http://covid.gilchrist.ca
Posts Media Videos Starter Packs
Pinned
COVID-19: Stats, Guides, Articles, Calculators

COVID-19 guides, useful info, stats, calculators & 600+ scientific articles: covid.gilchrist.ca

Ottawa, Canada weekly stats: covid.gilchrist.ca/Ottawa.html

Ontario, Canada weekly stats: covid.gilchrist.ca/Ontario.html

🧪🧵 1/
To learn how to see what variants are circulating in your area, please read this guide ( bsky.app/profile/jeff... ). 9/
How To: See what variants are circulating in your country/province/state

This thread will walk you through how to see which variants are circulating in your area using Mike Honey's visualization tool. 🧪🧵1/

On page unrolled version here ( pingthread.com/thread/16472... ).
To learn how to create a Sankey diagram of variants circulating in your area, please read this guide ( bsky.app/profile/jeff... ). 8/
How To: Create a Sankey diagram of variants circulating in your area

This thread will walk you through the process of using Mike Honey visualization tool to generate a Sankey diagram for your country, province or State. 🧪🧵1/

Unrolled version: threadreaderapp.com/thread/17454...
The SanKey diagram shows you variant lineages and you can follow back where that variant came from. 4/
It looks like PY.1.1.1 was trying to make a run for the top in mid-September but has decreased significantly. PY.1.1.1 is a descendant of LF.7 while the XFG variant is a recombination of LF.7 and LP.8.1.2 so they would share some mutations in common. 3/
Looking at specific variants, XFG.3 has dropped considerably and its descendent XFG.3.15 now takes first place at 6.2%, followed by XFG.3 at 5.2% and XFG.2 at 4.5%. 2/
** Ontario Variant Update (to Oct. 3, 2025) **

The XFG.* "Stratus" family is accounting for 83% while the next closest variant family NB.1.8.1.* "Nimbus" is 12.8% of sequenced genomes from COVID tests. 🧵1/
Health Canada has published Guidance for Indoor Air Quality Professionals, a practical, science-based resource for Indoor Air Quality (IAQ) experts, building operators, and employers.

Article ( ospe.on.ca/advocacy/pol... )

Guidelines ( www.canada.ca/en/health-ca... )
Policy Win: Health Canada Releases New IAQ Guidance
Health Canada announces new IAQ guidance, providing essential insights for improving indoor air quality and safety in buildings.
ospe.on.ca
Not sure where you are but some parts of the world including Japan are getting an updated Novavax LP.8.1 formula while USA is reusing the same formula as last year due to regulatory issues. Almost no reaction for Novavax for me too but mRNA this fall is using an updated LP.8.1 formula.
They are supposed to update their technical notes with more details in the next month. The last update was a year ago ( www.publichealthontario.ca/-/media/Data... ). 4/
www.publichealthontario.ca
I'm glad at least they didn't merge the 0-4 group into the other kids. It should already be broken down as <1 and 1-4 since that first year of life is vastly different for rates and impact on health than the kids 1-4 so it looks much less serious for infants when it gets averaged out. 3/
You can see that the latest data grouped the 5-11 and 12-17 into one age category of 5-17, the 18-39 & 40-59 categories have been grouped into 18-49 and the older adults are now grouped as 50-64, 65-74, and 75+. That is going to change the populate rates with the new groups. 2/
I found out for Ontario that "the Ministry of Health updated the age groups on the I9 bed census data collection forms" so some data in the Ontario Respiratory Virus Tool ( www.publichealthontario.ca/en/Data-and-... ) will be grouped differently starting in September. 🧵1/
It is interesting how companies are now highlighting just the fold increase in neutralizing antibodies since the absolute levels are many fold lower than what the original formula gave against the original variant and don't want to draw attention to that.
Reposted by Jeff Gilchrist
🧪 Can researchers stop AI from fabricating citations? Probably not entirely.

While models like GPT-5 are improving, hallucination is fundamental to their statistical design, shifting the burden onto users to build new intuitions for an AI age.
#MLSky #AcademicSky
Can researchers stop AI making up citations?
OpenAI’s GPT-5 hallucinates less than previous models do, but cutting hallucination completely might prove impossible.
www.nature.com
It depends on what part of the world you are in, but in Ontario during August 2024 the KP.3.1.1 variant (or some form of DeFLuQE mutation variant) would have been common.
To learn how to see what variants are circulating in your area, please read this guide ( bsky.app/profile/jeff... ). 18/
How To: See what variants are circulating in your country/province/state

This thread will walk you through how to see which variants are circulating in your area using Mike Honey's visualization tool. 🧪🧵1/

On page unrolled version here ( pingthread.com/thread/16472... ).
To learn how to create a Sankey diagram of variants circulating in your area, please read this guide ( bsky.app/profile/jeff... ). 17/
How To: Create a Sankey diagram of variants circulating in your area

This thread will walk you through the process of using Mike Honey visualization tool to generate a Sankey diagram for your country, province or State. 🧪🧵1/

Unrolled version: threadreaderapp.com/thread/17454...
The visualization tool for variants was created by @mikehoney.bsky.social and the source of this data can be found at ( app.powerbi.com/view?r=eyJrI... ). 16/
Power BI Report
Report powered by Power BI
app.powerbi.com