Elliot Schumacher retweeted
1/ We are publishing the first proof that AI can reason as well as clinicians in live clinical environments. Our AI diagnosed ~2400 patients as accurately as board certified clinicians in a real environment and was 30 percentage points better than Google’s equivalent system.
1/ Excited to announce our new preprint, Rare Disease Differential Diagnosis with Large Language Models at Scale: From Abdominal Actinomycosis to Wilson's Disease. We introduce a method that enhances rare disease diagnosis by combining LLMs with expert systems! @CuraiHQ 🧵
4/ RareScale uses an expert system and simulator to create structured cases, which are in turn fed into LLMs to create free-form dialogues. These dialogues allow us to train a rare disease candidate generator which can be used in combination with a black-box LLM.
5/ With RareScale, we saw a significant improvement, boosting Top-5 diagnostic accuracy up to 17% . Check out our full paper for details! arxiv.org/abs/2502.15069
1/ 🧵 Excited to share that we are presenting our paper, "MED-OMIT: Evaluation Metric for Omissions in Medical Summarization," at #GenAIforHealth and #ML4H2024! 🎉. Link : arxiv.org/abs/2311.08303
2/ By discretizing facts from source documents and assessing their significance to a downstream task (eg. diagnosis), MED-OMIT evaluates whether LLM-generated summaries retain all clinically important information for that task 🔎🩺
Great work Albert! Check out the paper at aclanthology.org/2024.naacl-… .
Had a wonderful time presenting my paper from my internship last year with @CuraiHQ at #NAACL2024!
Grateful for the opportunity to talk to the awesome and thoughtful people in the NLP community.
@elliotschu @anithakan @nairvarun18
Thrilled that @Albertyusun's work was accepted at #NAACL main conference! 🎉 Check out the paper arxiv.org/pdf/2304.14364.pdf and see the thread for details.
Excited to announce that my internship work with my awesome collaborators at @CuraiHQ from last year has been accepted to the main conference @naaclmeeting 2024😃
Check out our work building practical guardrail models for LLMs using a diverse synthetic data generation approach!
Elliot Schumacher retweeted
@CuraiHQ We are hiring researchers & engineers to work on highly impactful problems at the intersection of AI and healthcare to bring affordable care to everyone. tinyurl.com/2mv2camf
tinyurl.com/223es5vm
tinyurl.com/2vdcxe6f
Elliot Schumacher retweeted
1/ New preprint!🔔
Excited to share my internship work at @CuraiHQ.
Using model distillation techniques, we’ve developed an approach to create light-weight guardrail models that monitor the output of generative language models like GPT-4!
w/@nairvarun18 @elliotschu @anithakan
Elliot Schumacher retweeted
Latest from @curaiHQ: DERA - Dialog-Enabled Resolving Agents!
We (@elliotschu, @gtsomd, @anithakan) use the conversational ability of GPT-4 to create an interpretable forum for agents to communicate and iteratively improve output. Thread 👇
arxiv.org/abs/2303.17071
