How are predictors handled in the PLP package

[arani11]

I used a data driven approach and selected the observations (e.g., weight) and measurements (e.g., blood tests) as clinical domains to be included in my model. Is the default setting to include these as continuous measurements in the model? Can I see what variables are included in the "observations" and "measurements" categories?
Thanks

[jreps]

Note: the standard measurement features generated via "useMeasurementLongTerm = TRUE" are binary (indicating whether you had the measurement taken rather than the value), there may be a way to select the values for people who have values "useMeasurementValueAnyTimePrior = TRUE" but I do not recommend this since it does not handle missing values nicely, does not standardized units and I'm not sure how it handles multiple values. If I want measurement values, I generally write a custom covariate setting function such as https://github.com/ohdsi-studies/QRISKvalidation/blob/jenna/R/measurement_covariates.R

The easiest way the view what measurement features there are would be to view the covariateSummary in R and search for 'observation' or 'measurement' in the R viewer:

connectionDetails <- Eunomia::getEunomiaConnectionDetails()
Eunomia::createCohorts(connectionDetails)

modelDesign <- PatientLevelPrediction::createModelDesign(
targetId = 4,
outcomeId = 3,
restrictPlpDataSettings = PatientLevelPrediction::createRestrictPlpDataSettings(),
populationSettings = PatientLevelPrediction::createStudyPopulationSettings(
requireTimeAtRisk = FALSE,
riskWindowStart = 1,
riskWindowEnd = 365
),
covariateSettings = FeatureExtraction::createCovariateSettings(
useDemographicsGender = TRUE,
useDemographicsAgeGroup = TRUE,
useConditionGroupEraLongTerm = TRUE,
useDrugEraLongTerm = TRUE,
useMeasurementLongTerm = TRUE,
useProcedureOccurrenceLongTerm = TRUE
),
preprocessSettings = PatientLevelPrediction::createPreprocessSettings(),
modelSettings = PatientLevelPrediction::setLassoLogisticRegression(),
splitSettings = PatientLevelPrediction::createDefaultSplitSetting(
testFraction = 0.1, trainFraction = 0.9,
type = 'subject'
)
)

PatientLevelPrediction::runMultiplePlp(
databaseDetails = PatientLevelPrediction::createDatabaseDetails(
connectionDetails = connectionDetails,
cdmDatabaseSchema = 'main',
cdmDatabaseName = 'eunomia',
tempEmulationSchema = 'main',
cohortDatabaseSchema = 'main',
cohortTable = 'cohort',
outcomeDatabaseSchema = 'main',
outcomeTable = 'cohort',
cdmVersion = 5
),
modelDesignList = list(modelDesign),
saveDirectory = file.path(tempdir(), 'example_plp')
)

see folders

dir(file.path(tempdir(), 'example_plp'))

plpResult <- PatientLevelPrediction::loadPlpResult(file.path(tempdir(), 'example_plp', "Analysis_1", 'plpResult'))

open the data.frame in the R viewer

View(plpResult$covariateSummary)