Hejie Cui
@HennyJieCC
Followers
2K
Following
796
Media
13
Statuses
144
Postdoc Scholar @Stanford; CS PhD @EmoryUniversity; EECS Rising Star; previously @MSFTResearch, @AmazonScience; #machinelearning #datamining #AI4Health; She/her
Palo Alto
Joined November 2017
Introducing TIMER⌛️: a temporal instruction modeling and evaluation framework for longitudinal clinical records! 🏥📈. TIMER tackles challenges in processing longitudinal medical records—including temporal reasoning, multi-visit synthesis, and patient trajectory analysis. It.
1/🧵Introducing TIMER: Temporal Instruction Modeling and Evaluation for Longitudinal Clinical Records. When we evaluate LLMs for reasoning over longitudinal clinical records, can we leverage synthetic data generation to create scalable benchmarks and improve model performance?
0
8
34
RT @ddvd233: Excited to share our latest benchmark: CLIMB, where we built a solid data foundation for multimodal clinical models. With 4.51….
0
8
0
RT @jasonafries: 🎉 Excited to present our #ICLR2025 work—leveraging future medical outcomes to improve pretraining for prognostic vision mo….
0
7
0
RT @AlyssaUnell: Excited to present this work at ICLR's SynthData Workshop on Sunday April 27! Come through from 11:30-12:30 @ Peridot 202….
0
9
0
RT @StanfordMed: As artificial intelligence pervades health and medicine, Stanford Medicine researchers have developed a new evaluation fra….
0
9
0
We build 𝗠𝗲𝗱𝗛𝗘𝗟𝗠✨: a comprehensive benchmark evaluating AI on realistic clinical tasks that healthcare professionals perform daily instead of just medical exams.👩⚕️⚕️ . • Stanford HAI Blog: • Leaderboard:
1/🧵How do we know if AI is actually ready for healthcare? We built a benchmark, MedHELM, that tests LMs on real clinical tasks instead of just medical exams. #AIinHealthcare.Blog, GitHub, and link to leaderboard in thread!
0
5
33
RT @jasonafries: 🎉 We're thrilled to announce the general release of three de-identified, longitudinal EHR datasets from Stanford Medicine—….
0
26
0
Excited to present two papers at #NeurIPS 2024 this week! Stop by our posters if you're interested in LLMs, graphs, or health AI! I'm looking forward to reconnecting with familiar faces and meeting new friends! 🍻
0
8
54
RT @michigan_AI: Thrilled to unveil our⚡️Lightning Talks⚡️ for Michigan AI Symposium!.Join us to hear from:. @ZanaBucinca (@Harvard).Hejie….
0
4
0
RT @yuyinzhou_cs: 🎉 Extremely honored that our paper "A Preliminary Study of o1 in Medicine: Are We Getting Closer to an AI Doctor?" has be….
0
8
0
We have released a dataset of 80K clinician preference-aligned instruction-following examples and the instruction-tuned models on Hugging Face🤗: .Data: Models:
1
1
1
BioMed-VITAL incorporates clinician preferences in both data generation and selection stages, leading to more clinically relevant instruction data. The proposed data selection model 🧐, which distills a mixture of preferences, outperforms GPT-4 in matching human preferences.
1
1
1
Our work, BioMed-VITAL, which integrates clinician 👩⚕️preferences to generate and select high-quality instruction data for tuning biomedical multimodal AI models, has been accepted to the NeurIPS D&B Track. Paper: Website:
4
20
102
🧑💻 About to present our work on disease subtyping at #KDD2024. Please feel free to join us at 7:30 PM CET (Number 85).
0
2
29
RT @Abel0828: Happy to announce that both of our submissions have been accepted as Oral by the International Conference on Computational So….
0
2
0
RT @jasonafries: 🚀 Exciting News! 🚀. We're thrilled to announce that our longitudinal electronic health record (EHR) dataset, EHRSHOT, is n….
0
32
0
Exciting full-set assets (datasets, benchmarks, and models) for few-shot clinical prediction to evaluate medical foundation models for sample efficiency and task adaptation.🎯 .#MachineLearning #HealthcareML #FoundationModels.
Excited to announce the official full release of 👂💉EHRSHOT -- a dataset of 6,739 deidentified longitudinal EHRs for few-shot eval of foundation models!. 🌐Website: 🏥Dataset: 🤗Model:
0
0
10
RT @ylongqi: Attention academic friends! We are looking to fund creative research on the use of language models - small and large - in prod….
0
10
0