Skip to contents

Execution

a01_execution.Rmd

Executing the study

Now as we have initiated database connection and created the targetTable as well as the controlTable we are ready to execute the study.


################################################################################
#
# Execute
#
################################################################################

data = CohortContrast::CohortContrast(
  cdm,
  targetTable = targetTable,
  controlTable = controlTable,
  pathToResults = file.path(getwd(), "studies"),
  domainsIncluded = c(
    "Drug",
    "Condition",
    "Measurement",
    "Observation",
    "Procedure",
    "Visit",
    "Visit detail",
    "Death"
  ),
  prevalenceCutOff = 2.5,
  topK = FALSE, # Number of features to export
  presenceFilter = 0.2, # 0-1, percentage of people who must have the chosen feature present
  complementaryMappingTable = NULL, # Optional manual concept mapping table
  getSourceData = FALSE, # If true will generate summaries with source data as well
  runChi2YTests = TRUE,
  runLogitTests = FALSE,
  createOutputFiles = TRUE,
  complName = "LungCancer_1Y")

The parameters

There are multiple parameters we can tweak for different outcomes:

Mandatory:

cdm Connection to the database

targetTable Table for target cohort

controlTable Table for control cohort

pathToResults Path to the results folder, can be project’s working directory

domainsIncluded list of CDM domains to include, choose from Drug, Condition, Measurement, Observation, Procedure, Visit, Visit detail, Death

complName Name of the output study directory

Customization:

runChi2YTests boolean for running CHI2Y tests (chi-squared tests for two proportions with Yates continuity correction)

runLogitTests boolean for logit-tests on the prevalence, builds a model for predicting whether the patient is in target or control

getAllAbstractions boolean for creating abstractions’ levels for the imported data, this is useful when using GUI and exploring data

maximumAbstractionLevel Maximum level of abstraction allowed, if getAllAbstractions is TRUE, for hierarchy the concept_hierarchy table is used

getSourceData boolean for fetching source data, the data abstraction level which is used to map to OMOP CDM

prevalenceCutOff numeric or FALSE, if set, removes all of the concepts which are not present (in target) more than prevalenceCutOff times. Eg if set to 2, only concepts present double in target are exported.

topK numeric or FALSE, if set, keeps at maximum this number of features in the analysis. Maximum number of features exported.

presenceFilter numeric or FALSE, if set, removes all features represented by fewer target cohort subjects than the given percentage

complementaryMappingTable data frame or NULL. Mapping table for concept merges. Columns: CONCEPT_ID, CONCEPT_NAME, NEW_CONCEPT_ID, NEW_CONCEPT_NAME, ABSTRACTION_LEVEL, TYPE

numCores Number of cores to allocate to parallel processing, by default max number of cores - 1

createOutputFiles Boolean for creating output files, the default value is TRUE

runRemoveTemporalBias boolean for optional temporal-bias reduction step after main workflow

runAutomaticHierarchyCombineConcepts boolean for optional hierarchy-based post-processing

runAutomaticCorrelationCombineConcepts boolean for optional correlation-based post-processing

Notes:

When using the GUI prevalenceCutOff, presenceFilter can be changed on a slider.

The effect of runChi2YTests and runLogitTests can be toggled as a filter.

The function will output a study directory with complName, in this case LungCancer_1Y, inside pathToResults. The study directory contains parquet files (for example data_patients.parquet) and a metadata file metadata.json.

Reloading a saved study 

reloaded <- CohortContrast::loadCohortContrastStudy(
  studyName = "LungCancer_1Y",
  pathToResults = file.path(getwd(), "studies")
)