←
Nicholas Tatonetti5,019 posts
Nicholas Tatonetti avatar

Nicholas Tatonetti

@proftatonetti

Prof @CedarsSinai making drugs safer 1B data points at a time. @Stanford BMI PhD; @Columbia DBMI; @tatonetti.bsky.social; EiC BioData Mining; he/they 🏳️‍🌈

◷ Joined December 2008▧ 256 Photos and videos
Shivers avatar
⟳ Nicholas Tatonetti retweeted
Shivers@thinkingshivers·Oct 9

Instead of paying hundreds of thousands of dollars for a """doctor""" to take an X-Ray, you can get one for free from ChatGPT. My hand has been hurting like hell lately, but it turns out it's totally fine!

Post attachmentPost attachment
Emily Alsentzer avatar
⟳ Nicholas Tatonetti retweeted
Emily Alsentzer@Emily_Alsentzer·Aug 11

Clinical LLMs learn from EHR data not found in pretraining, but training them is expensive & they go stale as general models improve. We introduce CAPT, which steers a general model w/ a clinical model at inference (even w/ diff tokenizers!) Led by the stellar @SashaRonaghi 👇

Check out our paper “Training-Free Adaptation of New-Generation LLMs using Legacy Clinical Models” recently presented at @CHILconference

Boring_Business avatar
⟳ Nicholas Tatonetti retweeted
Boring_Business@BoringBiz_·Jun 5

CEOs walking into work after realizing that the AI tokens are going to cost more than the employees they fired

1:15
Nicholas Tatonetti avatar
Nicholas Tatonetti@proftatonetti·May 18

It's time again for the AMIA Translational Bioinformatics Year-in-Review This year: 2k papers triaged, 600+ reviewed. The final set is wild. Foundation models, multimodal bio, spatial everything, generative proteins, and some pure brain candy. Can't wait to present! #Amplify2026

Noah Zimmerman avatar
⟳ Nicholas Tatonetti retweeted
Noah Zimmerman@_noahzim_·May 13

I’m looking forward to hosting a conversation with leaders from @bmsnews and @TempusAI about a challenge that continues to come up in oncology: too many patients still face barriers to the testing and treatments that could help them most. Register here: tempus.co/4vSbDw2

Post attachment
Nicholas Tatonetti avatar
Nicholas Tatonetti@proftatonetti·Apr 22

OnSIDES v3.1.1 is out! 🎉 Updated with drug labels through April 2026 from all four sources (US, UK, EU, Japan): - 41K+ product labels - 6.9M drug–adverse event pairs - 4,097 ingredients Also fixes some bugs — if you're using that., please upgrade. github.com/tatonetti-lab/ons…

Nicholas Tatonetti avatar
Nicholas Tatonetti@proftatonetti·Mar 6

Some patterns stood out 1) the gap was largest in base Whisper with ~11 point higher word error rates for non-native speakers 2) errors clustered in clinically dense language 3) more biomedical terminology > higher error rates - medication/disease names were especially vulnerable

Nicholas Tatonetti avatar
Nicholas Tatonetti@proftatonetti·Mar 6

We evaluated Whisper/WhisperX Participants read clinical texts covering typical scenarios (colonoscopies or desc. clinical history) We include • native & non-native English speakers • with and without medical training Across models, errors were higher for non-native speakers

Nicholas Tatonetti avatar
Nicholas Tatonetti@proftatonetti·Mar 6

GI fellow Jamil Samaan noticed something odd when using AI ambient scribes and speech transcription tools Clinicians with non-native English accents appeared to confuse the model He and Yassi Fatapour decided to investigate, now out in npj Digital Medicine nature.com/articles/s41746-0…

Nicholas Tatonetti avatar
Nicholas Tatonetti@proftatonetti·Feb 6

who's going to point it their epic instance?

CLAUDE CODE but for HACKING its called shannon, you point it at website and it just... tries to break in... fully autonomous with no human needed i pointed it at a test app and it stole the entire user database, created admin accounts, and bypassed login, all by itself, in 90 minutes

Nicholas Tatonetti avatar
Nicholas Tatonetti@proftatonetti·Feb 5

New editorial in BioData Mining: A crisis of overconfidence in clinical AI led by @CedarsSinai PhD student, Jacob Berkowitz. Why sounding confident can be more dangerous than being inaccurate, and why calibration needs to be central in clinical AI. link.springer.com/article/10…

Demis Hassabis avatar
⟳ Nicholas Tatonetti retweeted
Demis Hassabis@demishassabis·Jan 29

AlphaGenome is our latest & most advanced genomics model published in @Nature today including making the model & weights available to academic researchers. Can’t wait to see what the research community will do with it. Congrats to the team on our newest front cover! #AI4Science

Our breakthrough AI model AlphaGenome is helping scientists understand our DNA, predict the molecular impact of genetic changes, and drive new biological discoveries. 🧬 Find out more in @Nature ↓ goo.gle/4bXlV6y

Post attachment
Eric Topol avatar
⟳ Nicholas Tatonetti retweeted
Eric Topol@EricTopol·Jan 29

The largest randomized trial of medical A.I. —Over 100,000 women in Sweden —radiologist + AI vs 2 radiologists, in follow-up —AI added led to 29% more cancer detected, 44% reduced workload, and —Less cancer dx in subsequent 2 years, and, when found, less aggressive thelancet.com/journals/lance…

Post attachment