dbt mimic omop is a free, open-source resource that converts the MIMIC-IV dataset to the Observational Medical Outcomes Partnership (OMOP) common data model (CDM) format on consumer level hardware. CDM approaches are increasingly adopted in both industry and academia due to the need for interoperability and reproducibility, including in clinical NLP tasks such as cohort selection, information extraction, and retrieval-augmented generation. The MIMIC-IV database is among the most widely used critical care research datasets, yet existing pipelines to transform it to OMOP depend on enterprise database infrastructure and complex orchestration, limiting accessibility for practitioners and resource-constrained researchers. We further integrate free-text clinical notes (195.6M clinical annotations) and chest radiographs into the OMOP note nlp and imaging extension tables, making all MIMIC-IV modalities (structured data, free-text, and imaging) accessible through a common data model. This resource generates a more comprehensive dataset than existing alternatives and is intended to be used to aid in system development, testing, and evaluation.

Retrieval-augmented generation (RAG) holds promise for clinical question answering over electronic health records (EHRs), but existing systems treat retrieval as an opaque subroutine, limiting auditability and reliability in patient care workflows. We introduce a deterministic multi-stage retrieval pipeline for longitudinal EHR question answering that decomposes retrieval into four distinct, ablated stages where each stage is instrumented with diagnostic metrics, making the flow of clinical evidence measurable and auditable at every step. Evaluated on a broad LLM-annotated cohort and an expert-annotated cardiovascular benchmark developed alongside clinicians from real ICU records, the full pipeline achieves 22-23% relative recall gain over a strong dense retrieval baseline across both cohorts, with consistent improvements in downstream answer quality. The pipeline’s deterministic and transparent design addresses a critical gap in clinical NLP: retrieval systems that clinicians and researchers can not only rely on, but inspect, audit, and build upon for real-world deployment.