Onyi Arah, MD, DSc, PhD
banner
oacarah.bsky.social
Onyi Arah, MD, DSc, PhD
@oacarah.bsky.social
Professor
Practical Causal Inference Lab Co-Director
**Views are mine**

#Epidemiology #EpiSky #CausalInference #CausalSky #PublicHealth #StatsSky #Stats #Medsky
Reposted by Onyi Arah, MD, DSc, PhD
DEBATE: "Is AI the future of health and social science?"

For those in London on 15 Dec 2025, don't miss this fun in person debate between me and David Bann sponsored by @ncrm.ac.uk!

Will the arguments change anyone's mind? 🤔

Sign up to attend IN PERSON: www.eventbrite.co.uk/e/is-ai-the-...
October 21, 2025 at 10:49 AM
Reposted by Onyi Arah, MD, DSc, PhD
Even as the anti-vax lobby erodes public trust, new evidence confirms that "immunizations against Covid-19, RSV, and influenza have shown consistent effectiveness & safety and are associated with a substantially reduced risk of hospitalization & severe disease"

www.nejm.org/doi/full/10....
October 29, 2025 at 11:15 PM
Reposted by Onyi Arah, MD, DSc, PhD
Respect to those who got here first.
October 13, 2025 at 6:09 PM
Familial confounding or measurement error? How to interpret findings from sibling and co-twin control studies - European Journal of Epidemiology
Epidemiological researchers often examine associations between risk factors and health outcomes in non-experimental designs. Observed associations may be causal or confounded by unmeasured factors. Sibling and co-twin control studies account for familial confounding by comparing exposure levels among siblings (or twins). If the exposure-outcome association is causal, the siblings should also differ regarding the outcome. However, such studies may sometimes introduce more bias than they alleviate. Measurement error in the exposure may bias results and lead to erroneous conclusions that truly causal exposure-outcome associations are confounded by familial factors. The current study used Monte Carlo simulations to examine bias due to measurement error in sibling control models when the observed exposure-outcome association is truly causal. The results showed that decreasing exposure reliability and increasing sibling-correlations in the exposure led to deflated exposure-outcome associations and inflated associations between the family mean of the exposure and the outcome. The risk of falsely concluding that causal associations were confounded was high in many situations. For example, when exposure reliability was 0.7 and the observed sibling-correlation was r = 0.4, about 30–90% of the samples (n = 2,000) provided results supporting a false conclusion of confounding, depending on how p-values were interpreted as evidence for a family effect on the outcome. The current results have practical importance for epidemiological researchers conducting or reviewing sibling and co-twin control studies and may improve our understanding of observed associations between risk factors and health outcomes. We have developed an app (SibSim) providing simulations of many situations not presented in this paper.
link.springer.com
October 12, 2025 at 3:20 PM
Reposted by Onyi Arah, MD, DSc, PhD
Reposted by Onyi Arah, MD, DSc, PhD
Breaking with the CDC, @acog.org reaffirms support for #COVID19 vaccination in pregnant people.

Here’s what to know.
ACOG Continues Recommending COVID-19 Vaccine During Pregnancy
This JAMA Medical News article discusses updated guidance on COVID-19 vaccination in pregnant people from the American College of Obstetricians and Gynecologists.
jamanetwork.com
September 12, 2025 at 5:46 PM
Reposted by Onyi Arah, MD, DSc, PhD
I am here to say vaccines are awesome because guess what Denmark is close to completely eliminating the two most dangerous HPV strains (serotype 16 and 18) and with the cervical cancer.I think that is bloody AMAZING!!!!
bit.ly/4pbHxjT
Denmark close to wiping out leading cancer-causing HPV strains after vaccine roll-out
A nationwide study suggests infections with human papillomavirus (HPV) types 16 and 18 have been virtually eliminated since vaccination began in 2008 – protecting even unvaccinated women.
bit.ly
September 8, 2025 at 9:52 PM
September 5, 2025 at 12:30 AM
Reposted by Onyi Arah, MD, DSc, PhD
💡A new paper by Elias Bareinboim and Drago Plecko underscores the intractability of ignorability assumptions commonly invoked in the potential outcomes framework, explains why structural causal models—explicitly grounded in well-defined causal mechanisms—are far easier to interpret. 1/2
August 24, 2025 at 1:19 PM
What a mess
September 4, 2025 at 5:34 PM
Reposted by Onyi Arah, MD, DSc, PhD
In response to the dismantling of the CDC, California, Oregon, and Washington have launched the West Coast Health Alliance, which will make their own science-based vaccine guidelines, and make sure vaccines are accessible to everyone.
September 3, 2025 at 6:18 PM
Reposted by Onyi Arah, MD, DSc, PhD
Not for nothing, whenever someone conflates search algorithms, LLMs, and whatever the fuck AI is, I send them this article.
September 2, 2025 at 6:48 AM
Reposted by Onyi Arah, MD, DSc, PhD
link 📈🤖
Challenges in Statistics: A Dozen Challenges in Causality and Causal Inference (Cinelli, Feller, Imbens et al) Causality and causal inference have emerged as core research areas at the interface of modern statistics and domains including biomedical sciences, social sciences, computer scie
August 26, 2025 at 4:17 PM
Every week is worse than the previous one
August 29, 2025 at 9:02 PM
Reposted by Onyi Arah, MD, DSc, PhD
"Guns killed more children than any other cause- more than cancer, more than car crashes- for 3 years in a row."- Washington Post
434 school shootings since the Columbine massacre, and our country still refuses to do what's needed to keep our children safe.
Yet we keep sending our kids to school
August 28, 2025 at 2:56 PM
This is bad
August 28, 2025 at 3:09 AM
The #AI hype is incredible
August 25, 2025 at 5:36 PM
On the Structural Basis of Conditional Ignorability

By Elias Bareinboim and Drago Plecko

#causalsky #statssky #stats #EpiSky

causalai.net/r120.pdf
causalai.net
August 24, 2025 at 6:15 PM
Reposted by Onyi Arah, MD, DSc, PhD
Thank you. To be clear, I did not merely “dismiss” AI as mid. My argument is that LLMs 1) are not intelligent 2) average textual responses in a way that is mid 3) most generous use cases are a revolution of middle skills that don’t produce anything ergo LLMs are in every demonstrable way mid.
August 23, 2025 at 2:29 PM