Turn as_plp_arrow() output into a plpData object — plp_arrow_to_plp

Assembles the cohort, outcome and covariate tables from as_plp_arrow() into the plpData structure that PatientLevelPrediction::createStudyPopulation() and runPlp() accept (and so DeepPLP). Observation-period arithmetic (daysFromObsStart, daysToObsEnd) and age / gender come from observation_period / person on con; rows with no observation period around their index date are dropped with a warning. Covariate ids are the 1-based positions in covariate_ref (analysis id 1), not FeatureExtraction's concept * 1000 + analysis.

Usage

plp_arrow_to_plp_data(
  plp_arrow,
  con,
  target_id = 1L,
  outcome_id = 1L,
  days_to_event = 1,
  cohort_end_days = 0
)

Arguments

plp_arrow: Result of as_plp_arrow() built with an outcome_col.
con: DuckDB connection holding observation_period and person.
target_id, outcome_id: Ids recorded in the cohort and outcome tables.
days_to_event: Day (relative to index) at which a positive label is placed (scalar or per row).
cohort_end_days: Days from index to cohort end.

Value

A plpData object.