Harvard Medical School · Mass General Brigham
Agentic AI for Real-World Healthcare
YLab develops agentic AI, clinical language models, and real-world evidence methods that turn complex health data into reliable tools for research and care.
Led by Prof. Jie Yang, with affiliations across Harvard Medical School, Brigham and Women's Hospital, the Broad Institute, Harvard Data Science Initiative, and the Kempner Institute.
Research
Our Research Areas
Our work spans agentic AI for healthcare, clinical language models and NLP, and pharmacoepidemiology using longitudinal real-world data.
Agentic AI for Healthcare
Building and evaluating AI agents that reason over clinical workflows, evidence, and tools with reliability and transparency.
Related publicationsClinical LLMs & NLP
Learning from clinical notes and EHR data to extract information, model patient trajectories, and support clinical research.
Software and resourcesPharmacoepidemiology & RWE
Combining EHR notes and insurance claims to generate scalable, trustworthy real-world evidence about medical products.
Related publicationsLatest updates
News
Service Dr. Yang joined the MGB Advisory Workgroup on AI in Research.
Publication BRIDGE was published in Nature Biomedical Engineering. Congratulations to Jiageng and Bowen. MGB news
Award Dr. Yang received the BWH Department of Medicine Chair's Research Award.
Honor Dr. Yang was elected a Fellow of the American College of Medical Informatics.
Funding YLab received two NIH R01 grants from NLM and NIA.
Open positions: one Postdoctoral Fellow in Agentic AI for Healthcare and one Research Specialist in AI for Healthcare.
Student Opportunities at YLab
We welcome PhD students, master's students, undergraduates, and collaborators from Harvard, MIT, and the greater Boston research community.
Eligible Harvard undergraduates may join through the KURE and KRANIUM programs.
Research environment
- Large-scale linked EHR and insurance claims data
- 20 advanced GPUs, including H100 and H200 systems
- Strong clinical, AI, and pharmacoepidemiology collaborations
Selected work
Featured publications
BRIDGE: Benchmarking Large Language Models for Understanding Real-World Clinical Practice Texts
A benchmark for testing whether language models can understand the practical, longitudinal reasoning documented in real-world clinical text.
Clinical Text Datasets for Medical Artificial Intelligence and Large Language Models
A systematic review and searchable landscape of clinical text datasets available for medical AI and language-model research.
Enhancing Postmarketing Surveillance of Medical Products with Large Language Models
A perspective on opportunities and safeguards for using large language models in postmarketing safety surveillance.
Research support
Funding and Institutional Support
We gratefully acknowledge support from the NIH, FDA, PCORI, and Mass General Brigham.
