Andrea Ganna
@andganna.bsky.social
1.5K followers 680 following 180 posts
Associate Prof in health data science @HiLIFE_helsinki @FIMM_UH - MGH/Harvard - Playing with all kind of data - http://dsgelab.org
Posts Media Videos Starter Packs
andganna.bsky.social
Thanks to the many biobanks involved!
And particularly Zhiyu Yang for leading this!
andganna.bsky.social
🧬💥 Do the genetics that make you develop a disease also help you survive it? Not much.

Our new study in Nature Genetics including 9 disease and 7 biobanks shows:

• Susceptibility variants ≠ survival
• PRSs for onset weak at predicting progression
• Lifespan PRS predicts survival better
andganna.bsky.social
comparing to one clinical prediction model is one way to think about this. the other is about the re-use of generated trajectories for multiple purposes in just one go: readmission rate, mortality, healthcare costs, disease risk and so on…
andganna.bsky.social
Thanks Michel! let’s chat. we want to have this pan-european.
andganna.bsky.social
One argument is that EHR are easily accessible why PRS are used only in research settings. so why bother if we can use already EHRs? showing that PRS provide complementary info makes them potentially useful. True especially for cancer where EHRs are very weak for disease prediction.
andganna.bsky.social
@detroiki.bsky.social and @tuomohartonen.bsky.social co-lead this project. We did it as part of INTERVENE. Thanks especially to Estonia Biobank colleagues! Always a pleasure to work together.

@fimm-uh.bsky.social
andganna.bsky.social
Check out @tuomohartonen.bsky.social blog post on our study, out today in Nature Genetics, comparing electronic health record-based risk scores with genetic polygenic scores, and how they work better together!

tuomohartonen.substack.com/p/medical-re...
andganna.bsky.social
Happy to see this online!

Having an extra sex chromosome really challenge the binary definition of sex based on XX and XY. So it is not only a medical, but also an important societal question.
andganna.bsky.social
Can you even access cosmos as university researcher?

I’m in general amazed how easy is to combine health data in US. It would impossible for a patient to opt out on the use of their data
andganna.bsky.social
Impressive paper by EPIC. They built a autoregressive transformer on 150M individuals and 115 billion
medical events.

Including: diagnoses, medications, lab values, procedure

Cons: model is not shared (and unlikely will be)

arxiv.org/pdf/2508.12104
andganna.bsky.social
Of course, if genetics is done routinely, there would be more public acceptance.
But the next frontier is: can an opt-out framework apply to biological specimens collected in clinic? Will this depend on which kind of -omics technique is used ? e.g. It might be ok for LDL but what about WGS?
andganna.bsky.social
thanks! to your question: Even now, if a genetic test is done in the clinic, researchers can access without consent under an opt-out framework. EHDS allows that (although countries might put some access restrictions specifically on genetics).
andganna.bsky.social
Conclusions: iPSYCH challenged the legal and ethical differences that exist between samples and data. This is a good thing. But, they should have envisioned an opt-out mechanism from the beginning, similar to what the European Health Data Space introduces for health data
andganna.bsky.social
I.e. they would not have the "key" to link samples to individuals. Now this fundamental rule has been broken, probably creating more damage than benefits.
andganna.bsky.social
Now iPSYCH researchers have been asked to inform participants they are in the study, so they can opt-out.

But this poses a new challenge: researchers have been originally granted access to these samples because there would not be any recontacting of participants.