Merge pull request #409 from OHDSI/develop

Develop

DESCRIPTION

@@ -2,8 +2,8 @@ Package: PatientLevelPrediction
 Type: Package
 Title: Developing patient level prediction using data in the OMOP Common Data
     Model
-Version: 6.3.3
-Date: 2023-05-15
+Version: 6.3.4
+Date: 2023-08-15
 Authors@R: c(
     person("Jenna", "Reps", email = "[email protected]", role = c("aut", "cre")),
     person("Martijn", "Schuemie", role = c("aut")),
@@ -39,7 +39,7 @@ Imports:
     polspline,
     pROC,
     PRROC,
-    reticulate (> 1.16),
+    reticulate (>= 1.30),
     rlang,
     SqlRender (>= 1.1.3),
     survival,

NAMESPACE

@@ -31,6 +31,8 @@ export(createPreprocessSettings)
 export(createRandomForestFeatureSelection)
 export(createRestrictPlpDataSettings)
 export(createSampleSettings)
+export(createSplineSettings)
+export(createStratifiedImputationSettings)
 export(createStudyPopulation)
 export(createStudyPopulationSettings)
 export(createTempModelLoc)

NEWS.md

@@ -1,3 +1,10 @@
+PatientLevelPrediction 6.3.4
+======================
+- added spline feature engineering 
+- added age/sex stratified imputation feature engineering
+- changed result table execution date types to varchar
+- updated covariateSummary to use feature engineering
+
 PatientLevelPrediction 6.3.3
 ======================
 - fixed bug introduced with new reticulate update in model saving to json tests

R/CovariateSummary.R

@@ -66,6 +66,28 @@ covariateSummary <- function(
     strata = strata
   )
 
+  # apply feature engineering
+  if(!is.null(featureEngineering)){
+
+    # create copy of covariateData
+    newCovariateData <- Andromeda::andromeda(
+      covariateRef = covariateData$covariateRef,
+      analysisRef = covariateData$analysisRef,
+      covariates = covariateData$covariates
+      )
+    covariateData <- newCovariateData
+
+    if(!is.null(featureEngineering$funct)){
+      featureEngineering <- list(featureEngineering)
+    }
+
+    for(fe in featureEngineering){
+      feSettings <- fe$settings
+      feSettings$trainData = list(covariateData = covariateData)
+      covariateData <- do.call(fe$funct, feSettings)$covariateData
+    }
+  }
+
   # make this run in parallel for big speed improvements..
   covariateSummariesPerStrata <- lapply(subsetList, 
     function(x){