NeurIPS'24 DB (Spotlight) | Instruction Tuning Large Language Models to Understand Electronic Health Records
-
Updated
Jul 17, 2026 - Jupyter Notebook
NeurIPS'24 DB (Spotlight) | Instruction Tuning Large Language Models to Understand Electronic Health Records
A lightweight clinical large language model (LLM) for cognitive decline detection from clinical notes. The project includes synthetic EHR data for rapid validation and experimentation. The paper was published in Journal of Biomedical Informatics: CD-Tron: Leveraging Large Clinical Language Models for Early Detection of Cognitive Decline from EHR
MCP server for FDA data: drug approvals, recalls, adverse events, 510(k) device clearances. Wraps openFDA.
Alan is an open-source structured system prompt, roughly 15,000 words, that guides a frontline health worker through an eye consultation. This study asks a simple question: how far does that scaffold travel when you move it from a frontier model down to a small model a rural clinic could actually run offline?
KAKAMEDLAB R Graph Gallery — an unofficial full-offline R chart Skill with upstream authorship preserved.
Research-grade, reproducible clinical-LLM auditing and mitigation workbench. Research use only.
To associate your repository with the clinical-llm topic, visit your repo's landing page and select "manage topics."