Suchi Saria
@suchisaria
AI prof/Endowed Chair @JohnsHopkins, Foundr @BayesianHealth, investor 15+ AI startups, ex-Stanford AI, MIT TR35, Sloan, @wef YGL, @darpa AI safety & Health AI
So incredibly proud.
The opportunity for AI to improve health outcomes is massive. But what’s been missing is rigorous peer-reviewed studies showing AI used in practice. Today we published 3 studies in Nature Medicine that show for the first time ever how AI saves lives. bayesianhealth.com/sepsis-researc…
Excited for us at @HopkinsEngineer to be part of this!
A great pleasure partnering with @CommerceGov @SecRaimondo and over 200 stakeholders to launch the U.S. Artificial Intelligence Safety Consortium with @NIST. The largest initiative of its kind, #AISIC will advance safe, secure, and trustworthy AI.
This is what rich people struggling looks like: wine parties w/o wine & billionaires sleeping on couches ;) wsj.com/lifestyle/trav… And apparently everybody is leading the global conversation on #AI! 😂 😂 😂
Could #AI help predict #sepsis? At 2:30pm join @DrMarcSiegel as he speaks with @suchisaria (AI prof/Endowed Chair @JohnsHopkins & Founder @BayesianHealth) and @withyouDrWu (Professor, director of CHSOR @JohnsHopkinsSPH) about their research. Stream here: sxm.app.link/DoctorRadio
AI monitoring is key to responsible deployment. Our #ICML2025 paper develops approaches for 3 main goals: 1) *Adapting* to mild data shifts 2) *Quickly Detecting* harmful shifts 3) *Diagnosing* cause of degradation 🧵w/ @xinghan0 @anqi_liu33 @suchisaria arxiv.org/abs/2505.04608
🚨With an awesome group of collaborators from health system & federal agencies, @Jean_J_Feng and I are embarking on an ambitious multi-year project to develop tools for monitoring & updating clinical AI algorithms, with the aim of informing how we accelerate & scale responsible…
Awesome highlights of the important research being done by @MathiasUnberath and @suchisaria! #ResearchSavesLives
From diagnosing infections with a smartphone to training surgical robots, Johns Hopkins researchers are using AI to create real-world impact. hub.jhu.edu/research-saves…
“For rare disease, the journey is 7 years before diagnosis” - Dr. Janet Woodcock wow.🤯
Hopkins Engineering faculty Rama Chellappa, @suchisaria, and Axel Krieger talk AI healthcare with @ScienceNews!
AI could transform health care, but the stakes are high. sciencenews.org/article/ai-hea…
When do AI explanations actually help, & promote appropriate trust? Spoiler, via prospective, multisite @radiology_rsna study of 220 doctors: *How* AI explains its advice has big impacts on doctors’ diagnostic performance and trust in AI--even if they *don’t realize it*! 🧵1/
Paper 🧵! w/ @samuel_stanton_ @anqi_liu33 @suchisaria #ICML2024 arxiv.org/abs/2405.06627 We study 2 Qs: 1) Can AI uncertainty quant. via #ConformalPrediction extend to any data distribution? 2) Are there practical CP algorithms for AI/ML agents 🤖w/ feedback-loop data shifts? 1/
Thank you 🙏 cc @DrewPrinster
This is a monumental breakthrough in the realm of Conformal Prediction, akin to discovering another universe in physics! 🔥🔥🔥🔥🔥 A groundbreaking new paper from John Hopkins University and Genentech titled "Conformal Validity Guarantees Exist for Any Data Distribution" by…
Had a (positive) “deer-in-the-headlights” moment today when scrolled to these kind & generous words from @predict_addict on our new paper w/ @samuel_stanton_ @anqi_liu33 @suchisaria! 😆arxiv.org/abs/2405.06627 Humbled & grateful to share this work. Full paper summary thread soon!
This is a monumental breakthrough in the realm of Conformal Prediction, akin to discovering another universe in physics! 🔥🔥🔥🔥🔥 A groundbreaking new paper from John Hopkins University and Genentech titled "Conformal Validity Guarantees Exist for Any Data Distribution" by…
‘Nuf said.
Broke out some S tier @scale_AI swag (from the vault)
Johns Hopkins and Columbia University computer scientists have proposed a method to enhance the robustness of AI models used in medical text analysis. The project is part of an ongoing effort to develop an AI safety framework for healthcare applications. bit.ly/48yyPCx
Computer scientists from @JohnsHopkins @HopkinsEngineer (featuring @suchisaria) and @Columbia combat the inaccurate correlations that #AI learns from text data. Learn more: hub.jhu.edu/2024/02/20/art…
#1stworldproblems #1stworldjokes After years, I recently got downgraded from United Global Services to 1K. The air hostesses think they’re being nice when they greet me — “You’re a 1K member! Thank you for your loyalty.” It mostly hurts, feels like they’re rubbing it in! 😂😂
Will your LLM model generalize over time or across sites? How do we improve robustness? This is a key issue LLMs struggle with and especially critical as we apply these models to healthcare applications. 👇 New clever technique to improve LLM robustness and generalization.…
