Skip to contents

Run the PCORnet -> OMOP CDM v5.4 ETL (R implementation)

Source: R/etl_pcornet.R
etl_pcornet.Rd

Mirrors python/omop_etl/build_omop_cdm.py table-for-table. Both read the same PCORnet-format source directory, share the same schema and concept-resolution logic, and generate 100% deterministic surrogate keys and OHDSI v5.4 tables.

Usage

etl_pcornet(
  source_dir,
  db_path = "omop_cdm.duckdb",
  central_vocab = NULL,
  site_id = NULL,
  site_anon = NULL,
  site_name = NULL,
  disambiguate_patids = FALSE,
  auto_resources = FALSE,
  memory_limit = NULL,
  threads = NULL,
  temp_dir = NULL
)

Arguments

source_dir

Directory containing PCORnet-format CSVs (demographic.csv, encounter.csv, diagnosis.csv, procedures.csv, lab_result_cm.csv, prescribing.csv, provider.csv, optionally vital.csv, facility.csv, and optionally death.csv), matched case-insensitively. Any table whose source file isn't found is skipped.

db_path

Path to the DuckDB database file (schema + vocabulary must already exist).

central_vocab

Optional path to an existing central vocabulary DuckDB database to attach in read-only mode with zero-copy views.

site_id

Optional numeric site identifier (e.g. 1 to 5).

site_anon

Optional publication pseudonym (e.g. "Site A").

site_name

Optional institutional name or internal descriptor (e.g. "Hospital System A").

disambiguate_patids

Logical. If TRUE, formats person_source_value as src.PATID || '-' || site_id and derives surrogate keys via pcornet_id(src.PATID || '-' || site_id).

auto_resources

Logical. If TRUE, size DuckDB to the host with configure_resources() (75% of available RAM, best spill volume, cores - 1) before loading; memory_limit, threads and temp_dir still win over the detected values.

memory_limit

Optional DuckDB memory_limit, e.g. "16GB".

threads

Optional DuckDB worker thread count.

temp_dir

Optional DuckDB disk-spill directory.

Value

Invisibly, db_path.