Coursera Spring Sale
40% Off Coursera Plus Annual!
Grab it
Learn how to develop a Retrieval-Augmented Generation (RAG)-informed Large Language Model workflow specifically designed for observational health research using the Julia programming language ecosystem. Explore an innovative approach that combines domain-specific query languages with advanced AI techniques to simplify complex patient dataset investigations stored in the OMOP Common Data Model (OMOP CDM). Discover how the JuliaHealth community's specialized tools, particularly FunSQL.jl's domain-specific language, can abstract SQL complexities and translate high-level query expressions into executable commands for pharmacovigilance, public health surveillance, and health economics research. Examine the development of a comprehensive knowledge corpus drawn from FunSQL.jl, OMOPCDMCohortCreator.jl, and OMOP CDM standards that serves as the foundation for contextual understanding in automated query generation. Understand the implementation of agentic workflows using JuliaGenAI community tools like PromptingTools.jl and RAGTools.jl, which enable dynamic, iterative processes where AI systems actively engage with data and human feedback to continuously refine outputs. Investigate the evaluation methodology for state-of-the-art local LLMs based on inference times, accuracy, and adherence to Julia ecosystem best practices. Learn about hybrid RAG architecture design incorporating advanced embedding models with optimized dimensions, vector databases, and traditional techniques like BM25 to complement embedding-based methods. Gain insights into how this workflow bridges the gap between automated query generation and human expert validation, ensuring semantically correct and contextually meaningful queries for large-scale health data analysis across diverse clinical datasets.
