MABadge

[MABadge @MABadge]

To healthcare AI practitioners: make calculated decisions about how far to deploy your models. Our newest study shows how to improve language model generalizability and make data-driven predictions of model performance on new (unlabeled) datasets! https://t.co/oNCxWRu9EK

[Ben Glicksberg @BenGlicksberg]

Many incredible studies show the power of EHR + biobanks for discovery. Check out our orthogonal approach to identify putative cardio-metabolic drug targets via loss-of-function variants across multiple cohorts and modalities!

https://t.co/69ywdIsfPY

#Bioinformatics #CompBio

[MABadge @MABadge]

Deep Learning Exploits Clinical Reasoning to Predict Hip Fracture in X-rays. Case-control matching blunts performance.
https://t.co/IObB6eVJiZ

[MABadge @MABadge]

@SimonDeDeo As an MD/PhD student at Verily (Google) Life Sciences, I can attest that they do cool (and impactful) research. I've appreciated the strengths and weakness of academia and industry, and while academic ideals are lofty, in practice the scale does not tip as cleanly as you suggest.

[MABadge @MABadge]

"It's not everyday that I hear something I haven't thought of before" #neurosciencechair

[MABadge @MABadge]

@TweetlessBrian whatever you send me part files for. And the parts for an xy coordinate laser cutter @facialnerve415 wants to make.

[MABadge @MABadge]

Y axis done for reprappro huxley!

[MABadge @MABadge]

@schleyman4 @nickaknudson good guess.