Missing data

[arani11]

Using the PLP package is it possible to do the following (as per TRIPOD+AI guidelines):

1. For each predictor included in the model report the number of missing values
2. Do multiple imputation for missing data

I developed a model using the PLP package a year ago and now writing it up using the TRIPOD+AI guidelines and wanted to do the above, if possible, to adhere to these guidelines.

Thanks.

[jreps]

Many of the features are designed so that there is no missingness (i.e., the features are generally 'code X recorded in prior year' which is 1 if the patient has the code X recorded and 0 otherwise). If you use measurements, then you can have missingness.

Egill has added a few multiple imputation methods:
PatientLevelPrediction::createIterativeImputer()
PatientLevelPrediction::createSklearnIterativeImputer()

You can use this by specifying it as the feature engineering:
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
),
featureEngineeringSettings = PatientLevelPrediction::createIterativeImputer(),
preprocessSettings = PatientLevelPrediction::createPreprocessSettings(),
modelSettings = PatientLevelPrediction::setLassoLogisticRegression(),
splitSettings = PatientLevelPrediction::createDefaultSplitSetting(
testFraction = 0.1, trainFraction = 0.9,
type = 'subject'
)
)