R/modelbuilder.h.R
In sbalci/ClinicoPathJamoviModule: Analysis for Clinicopathological Research
Defines functions
modelbuilder
# This file is automatically generated, you probably don't want to edit this
modelbuilderOptions <- if (requireNamespace("jmvcore", quietly=TRUE)) R6::R6Class(
"modelbuilderOptions",
inherit = jmvcore::Options,
public = list(
initialize = function(
outcome = NULL,
outcomePositive = NULL,
splitData = TRUE,
randomSeed = 123,
buildBasicModel = TRUE,
basicPredictors = NULL,
basicModelName = "basic_model",
buildEnhancedModel = FALSE,
enhancedPredictors = NULL,
enhancedModelName = "enhanced_model",
buildBiomarkerModel = FALSE,
biomarkerPredictors = NULL,
biomarkerModelName = "biomarker_model",
buildCustomModel = FALSE,
customPredictors = NULL,
customModelName = "custom_model",
includeInteractions = FALSE,
interactionTerms = "",
useStepwise = FALSE,
stepwiseDirection = "both",
selectionCriterion = "aic",
transformVariables = FALSE,
transformMethod = "log",
missingDataMethod = "complete_cases",
imputationSets = 5,
crossValidation = FALSE,
cvFolds = 5,
bootstrapValidation = FALSE,
bootstrapReps = 1000,
showModelSummary = TRUE,
showPerformanceMetrics = TRUE,
showCalibrationPlots = TRUE,
showROCCurves = TRUE,
compareModels = TRUE,
createPredictions = TRUE,
exportForDCA = TRUE,
calculateNRI = FALSE,
nriThresholds = "0.05, 0.10, 0.20",
calculateIDI = FALSE,
penalizedRegression = FALSE,
penaltyType = "lasso",
generateRiskScore = FALSE,
riskScorePoints = "simple", ...) {
super$initialize(
package="ClinicoPath",
name="modelbuilder",
requiresData=TRUE,
...)
private$..outcome <- jmvcore::OptionVariable$new(
"outcome",
outcome,
suggested=list(
"nominal"),
permitted=list(
"factor"))
private$..outcomePositive <- jmvcore::OptionLevel$new(
"outcomePositive",
outcomePositive,
variable="(outcome)")
private$..splitData <- jmvcore::OptionBool$new(
"splitData",
splitData,
default=TRUE)
private$..randomSeed <- jmvcore::OptionNumber$new(
"randomSeed",
randomSeed,
default=123,
min=1,
max=99999)
private$..buildBasicModel <- jmvcore::OptionBool$new(
"buildBasicModel",
buildBasicModel,
default=TRUE)
private$..basicPredictors <- jmvcore::OptionVariables$new(
"basicPredictors",
basicPredictors,
suggested=list(
"continuous",
"ordinal",
"nominal"),
permitted=list(
"numeric",
"factor"))
private$..basicModelName <- jmvcore::OptionString$new(
"basicModelName",
basicModelName,
default="basic_model")
private$..buildEnhancedModel <- jmvcore::OptionBool$new(
"buildEnhancedModel",
buildEnhancedModel,
default=FALSE)
private$..enhancedPredictors <- jmvcore::OptionVariables$new(
"enhancedPredictors",
enhancedPredictors,
suggested=list(
"continuous",
"ordinal",
"nominal"),
permitted=list(
"numeric",
"factor"))
private$..enhancedModelName <- jmvcore::OptionString$new(
"enhancedModelName",
enhancedModelName,
default="enhanced_model")
private$..buildBiomarkerModel <- jmvcore::OptionBool$new(
"buildBiomarkerModel",
buildBiomarkerModel,
default=FALSE)
private$..biomarkerPredictors <- jmvcore::OptionVariables$new(
"biomarkerPredictors",
biomarkerPredictors,
suggested=list(
"continuous",
"ordinal",
"nominal"),
permitted=list(
"numeric",
"factor"))
private$..biomarkerModelName <- jmvcore::OptionString$new(
"biomarkerModelName",
biomarkerModelName,
default="biomarker_model")
private$..buildCustomModel <- jmvcore::OptionBool$new(
"buildCustomModel",
buildCustomModel,
default=FALSE)
private$..customPredictors <- jmvcore::OptionVariables$new(
"customPredictors",
customPredictors,
suggested=list(
"continuous",
"ordinal",
"nominal"),
permitted=list(
"numeric",
"factor"))
private$..customModelName <- jmvcore::OptionString$new(
"customModelName",
customModelName,
default="custom_model")
private$..includeInteractions <- jmvcore::OptionBool$new(
"includeInteractions",
includeInteractions,
default=FALSE)
private$..interactionTerms <- jmvcore::OptionString$new(
"interactionTerms",
interactionTerms,
default="")
private$..useStepwise <- jmvcore::OptionBool$new(
"useStepwise",
useStepwise,
default=FALSE)
private$..stepwiseDirection <- jmvcore::OptionList$new(
"stepwiseDirection",
stepwiseDirection,
options=list(
"both",
"forward",
"backward"),
default="both")
private$..selectionCriterion <- jmvcore::OptionList$new(
"selectionCriterion",
selectionCriterion,
options=list(
"aic",
"bic"),
default="aic")
private$..transformVariables <- jmvcore::OptionBool$new(
"transformVariables",
transformVariables,
default=FALSE)
private$..transformMethod <- jmvcore::OptionList$new(
"transformMethod",
transformMethod,
options=list(
"log",
"polynomial",
"spline"),
default="log")
private$..missingDataMethod <- jmvcore::OptionList$new(
"missingDataMethod",
missingDataMethod,
options=list(
"complete_cases",
"mean_imputation",
"multiple_imputation",
"exclude_missing"),
default="complete_cases")
private$..imputationSets <- jmvcore::OptionNumber$new(
"imputationSets",
imputationSets,
default=5,
min=3,
max=20)
private$..crossValidation <- jmvcore::OptionBool$new(
"crossValidation",
crossValidation,
default=FALSE)
private$..cvFolds <- jmvcore::OptionNumber$new(
"cvFolds",
cvFolds,
default=5,
min=3,
max=10)
private$..bootstrapValidation <- jmvcore::OptionBool$new(
"bootstrapValidation",
bootstrapValidation,
default=FALSE)
private$..bootstrapReps <- jmvcore::OptionNumber$new(
"bootstrapReps",
bootstrapReps,
default=1000,
min=100,
max=5000)
private$..showModelSummary <- jmvcore::OptionBool$new(
"showModelSummary",
showModelSummary,
default=TRUE)
private$..showPerformanceMetrics <- jmvcore::OptionBool$new(
"showPerformanceMetrics",
showPerformanceMetrics,
default=TRUE)
private$..showCalibrationPlots <- jmvcore::OptionBool$new(
"showCalibrationPlots",
showCalibrationPlots,
default=TRUE)
private$..showROCCurves <- jmvcore::OptionBool$new(
"showROCCurves",
showROCCurves,
default=TRUE)
private$..compareModels <- jmvcore::OptionBool$new(
"compareModels",
compareModels,
default=TRUE)
private$..createPredictions <- jmvcore::OptionBool$new(
"createPredictions",
createPredictions,
default=TRUE)
private$..exportForDCA <- jmvcore::OptionBool$new(
"exportForDCA",
exportForDCA,
default=TRUE)
private$..calculateNRI <- jmvcore::OptionBool$new(
"calculateNRI",
calculateNRI,
default=FALSE)
private$..nriThresholds <- jmvcore::OptionString$new(
"nriThresholds",
nriThresholds,
default="0.05, 0.10, 0.20")
private$..calculateIDI <- jmvcore::OptionBool$new(
"calculateIDI",
calculateIDI,
default=FALSE)
private$..penalizedRegression <- jmvcore::OptionBool$new(
"penalizedRegression",
penalizedRegression,
default=FALSE)
private$..penaltyType <- jmvcore::OptionList$new(
"penaltyType",
penaltyType,
options=list(
"lasso",
"ridge",
"elastic_net"),
default="lasso")
private$..generateRiskScore <- jmvcore::OptionBool$new(
"generateRiskScore",
generateRiskScore,
default=FALSE)
private$..riskScorePoints <- jmvcore::OptionList$new(
"riskScorePoints",
riskScorePoints,
options=list(
"framingham",
"simple",
"deciles"),
default="simple")
self$.addOption(private$..outcome)
self$.addOption(private$..outcomePositive)
self$.addOption(private$..splitData)
self$.addOption(private$..randomSeed)
self$.addOption(private$..buildBasicModel)
self$.addOption(private$..basicPredictors)
self$.addOption(private$..basicModelName)
self$.addOption(private$..buildEnhancedModel)
self$.addOption(private$..enhancedPredictors)
self$.addOption(private$..enhancedModelName)
self$.addOption(private$..buildBiomarkerModel)
self$.addOption(private$..biomarkerPredictors)
self$.addOption(private$..biomarkerModelName)
self$.addOption(private$..buildCustomModel)
self$.addOption(private$..customPredictors)
self$.addOption(private$..customModelName)
self$.addOption(private$..includeInteractions)
self$.addOption(private$..interactionTerms)
self$.addOption(private$..useStepwise)
self$.addOption(private$..stepwiseDirection)
self$.addOption(private$..selectionCriterion)
self$.addOption(private$..transformVariables)
self$.addOption(private$..transformMethod)
self$.addOption(private$..missingDataMethod)
self$.addOption(private$..imputationSets)
self$.addOption(private$..crossValidation)
self$.addOption(private$..cvFolds)
self$.addOption(private$..bootstrapValidation)
self$.addOption(private$..bootstrapReps)
self$.addOption(private$..showModelSummary)
self$.addOption(private$..showPerformanceMetrics)
self$.addOption(private$..showCalibrationPlots)
self$.addOption(private$..showROCCurves)
self$.addOption(private$..compareModels)
self$.addOption(private$..createPredictions)
self$.addOption(private$..exportForDCA)
self$.addOption(private$..calculateNRI)
self$.addOption(private$..nriThresholds)
self$.addOption(private$..calculateIDI)
self$.addOption(private$..penalizedRegression)
self$.addOption(private$..penaltyType)
self$.addOption(private$..generateRiskScore)
self$.addOption(private$..riskScorePoints)
}),
active = list(
outcome = function() private$..outcome$value,
outcomePositive = function() private$..outcomePositive$value,
splitData = function() private$..splitData$value,
randomSeed = function() private$..randomSeed$value,
buildBasicModel = function() private$..buildBasicModel$value,
basicPredictors = function() private$..basicPredictors$value,
basicModelName = function() private$..basicModelName$value,
buildEnhancedModel = function() private$..buildEnhancedModel$value,
enhancedPredictors = function() private$..enhancedPredictors$value,
enhancedModelName = function() private$..enhancedModelName$value,
buildBiomarkerModel = function() private$..buildBiomarkerModel$value,
biomarkerPredictors = function() private$..biomarkerPredictors$value,
biomarkerModelName = function() private$..biomarkerModelName$value,
buildCustomModel = function() private$..buildCustomModel$value,
customPredictors = function() private$..customPredictors$value,
customModelName = function() private$..customModelName$value,
includeInteractions = function() private$..includeInteractions$value,
interactionTerms = function() private$..interactionTerms$value,
useStepwise = function() private$..useStepwise$value,
stepwiseDirection = function() private$..stepwiseDirection$value,
selectionCriterion = function() private$..selectionCriterion$value,
transformVariables = function() private$..transformVariables$value,
transformMethod = function() private$..transformMethod$value,
missingDataMethod = function() private$..missingDataMethod$value,
imputationSets = function() private$..imputationSets$value,
crossValidation = function() private$..crossValidation$value,
cvFolds = function() private$..cvFolds$value,
bootstrapValidation = function() private$..bootstrapValidation$value,
bootstrapReps = function() private$..bootstrapReps$value,
showModelSummary = function() private$..showModelSummary$value,
showPerformanceMetrics = function() private$..showPerformanceMetrics$value,
showCalibrationPlots = function() private$..showCalibrationPlots$value,
showROCCurves = function() private$..showROCCurves$value,
compareModels = function() private$..compareModels$value,
createPredictions = function() private$..createPredictions$value,
exportForDCA = function() private$..exportForDCA$value,
calculateNRI = function() private$..calculateNRI$value,
nriThresholds = function() private$..nriThresholds$value,
calculateIDI = function() private$..calculateIDI$value,
penalizedRegression = function() private$..penalizedRegression$value,
penaltyType = function() private$..penaltyType$value,
generateRiskScore = function() private$..generateRiskScore$value,
riskScorePoints = function() private$..riskScorePoints$value),
private = list(
..outcome = NA,
..outcomePositive = NA,
..splitData = NA,
..randomSeed = NA,
..buildBasicModel = NA,
..basicPredictors = NA,
..basicModelName = NA,
..buildEnhancedModel = NA,
..enhancedPredictors = NA,
..enhancedModelName = NA,
..buildBiomarkerModel = NA,
..biomarkerPredictors = NA,
..biomarkerModelName = NA,
..buildCustomModel = NA,
..customPredictors = NA,
..customModelName = NA,
..includeInteractions = NA,
..interactionTerms = NA,
..useStepwise = NA,
..stepwiseDirection = NA,
..selectionCriterion = NA,
..transformVariables = NA,
..transformMethod = NA,
..missingDataMethod = NA,
..imputationSets = NA,
..crossValidation = NA,
..cvFolds = NA,
..bootstrapValidation = NA,
..bootstrapReps = NA,
..showModelSummary = NA,
..showPerformanceMetrics = NA,
..showCalibrationPlots = NA,
..showROCCurves = NA,
..compareModels = NA,
..createPredictions = NA,
..exportForDCA = NA,
..calculateNRI = NA,
..nriThresholds = NA,
..calculateIDI = NA,
..penalizedRegression = NA,
..penaltyType = NA,
..generateRiskScore = NA,
..riskScorePoints = NA)
)
modelbuilderResults <- if (requireNamespace("jmvcore", quietly=TRUE)) R6::R6Class(
"modelbuilderResults",
inherit = jmvcore::Group,
active = list(
instructions = function() private$.items[["instructions"]],
dataSummary = function() private$.items[["dataSummary"]],
dcaReadyMessage = function() private$.items[["dcaReadyMessage"]],
basicModelSummary = function() private$.items[["basicModelSummary"]],
enhancedModelSummary = function() private$.items[["enhancedModelSummary"]],
biomarkerModelSummary = function() private$.items[["biomarkerModelSummary"]],
customModelSummary = function() private$.items[["customModelSummary"]],
modelComparisonTable = function() private$.items[["modelComparisonTable"]],
validationResults = function() private$.items[["validationResults"]],
nriTable = function() private$.items[["nriTable"]],
idiTable = function() private$.items[["idiTable"]],
riskScoreTable = function() private$.items[["riskScoreTable"]],
rocCurvesPlot = function() private$.items[["rocCurvesPlot"]],
calibrationPlotsArray = function() private$.items[["calibrationPlotsArray"]],
modelComparisonPlot = function() private$.items[["modelComparisonPlot"]],
validationPlot = function() private$.items[["validationPlot"]],
dcaPreparationSummary = function() private$.items[["dcaPreparationSummary"]]),
private = list(),
public=list(
initialize=function(options) {
super$initialize(
options=options,
name="",
title="Prediction Model Builder",
refs=list(
"glmnet",
"rms",
"pROC",
"mice",
"VIM",
"ClinicoPathJamoviModule"))
self$add(jmvcore::Html$new(
options=options,
name="instructions",
title="Instructions",
visible=TRUE))
self$add(jmvcore::Html$new(
options=options,
name="dataSummary",
title="Data Summary",
visible=TRUE))
self$add(jmvcore::Html$new(
options=options,
name="dcaReadyMessage",
title="Ready for Decision Curve Analysis",
visible="(createPredictions)"))
self$add(jmvcore::Table$new(
options=options,
name="basicModelSummary",
title="Basic Clinical Model",
visible="(buildBasicModel && showModelSummary)",
columns=list(
list(
`name`="term",
`title`="Variable",
`type`="text"),
list(
`name`="estimate",
`title`="Coefficient",
`type`="number",
`format`="zto"),
list(
`name`="std_error",
`title`="Std. Error",
`type`="number",
`format`="zto"),
list(
`name`="z_value",
`title`="z-value",
`type`="number",
`format`="zto"),
list(
`name`="p_value",
`title`="p-value",
`type`="number",
`format`="zto,pvalue"),
list(
`name`="odds_ratio",
`title`="Odds Ratio",
`type`="number",
`format`="zto"),
list(
`name`="ci_lower",
`title`="OR 95% CI Lower",
`type`="number",
`format`="zto"),
list(
`name`="ci_upper",
`title`="OR 95% CI Upper",
`type`="number",
`format`="zto")),
clearWith=list(
"basicPredictors",
"outcome",
"outcomePositive")))
self$add(jmvcore::Table$new(
options=options,
name="enhancedModelSummary",
title="Enhanced Clinical Model",
visible="(buildEnhancedModel && showModelSummary)",
columns=list(
list(
`name`="term",
`title`="Variable",
`type`="text"),
list(
`name`="estimate",
`title`="Coefficient",
`type`="number",
`format`="zto"),
list(
`name`="std_error",
`title`="Std. Error",
`type`="number",
`format`="zto"),
list(
`name`="z_value",
`title`="z-value",
`type`="number",
`format`="zto"),
list(
`name`="p_value",
`title`="p-value",
`type`="number",
`format`="zto,pvalue"),
list(
`name`="odds_ratio",
`title`="Odds Ratio",
`type`="number",
`format`="zto"),
list(
`name`="ci_lower",
`title`="OR 95% CI Lower",
`type`="number",
`format`="zto"),
list(
`name`="ci_upper",
`title`="OR 95% CI Upper",
`type`="number",
`format`="zto")),
clearWith=list(
"enhancedPredictors",
"outcome",
"outcomePositive")))
self$add(jmvcore::Table$new(
options=options,
name="biomarkerModelSummary",
title="Biomarker Model",
visible="(buildBiomarkerModel && showModelSummary)",
columns=list(
list(
`name`="term",
`title`="Variable",
`type`="text"),
list(
`name`="estimate",
`title`="Coefficient",
`type`="number",
`format`="zto"),
list(
`name`="std_error",
`title`="Std. Error",
`type`="number",
`format`="zto"),
list(
`name`="z_value",
`title`="z-value",
`type`="number",
`format`="zto"),
list(
`name`="p_value",
`title`="p-value",
`type`="number",
`format`="zto,pvalue"),
list(
`name`="odds_ratio",
`title`="Odds Ratio",
`type`="number",
`format`="zto"),
list(
`name`="ci_lower",
`title`="OR 95% CI Lower",
`type`="number",
`format`="zto"),
list(
`name`="ci_upper",
`title`="OR 95% CI Upper",
`type`="number",
`format`="zto")),
clearWith=list(
"biomarkerPredictors",
"outcome",
"outcomePositive")))
self$add(jmvcore::Table$new(
options=options,
name="customModelSummary",
title="Custom Model",
visible="(buildCustomModel && showModelSummary)",
columns=list(
list(
`name`="term",
`title`="Variable",
`type`="text"),
list(
`name`="estimate",
`title`="Coefficient",
`type`="number",
`format`="zto"),
list(
`name`="std_error",
`title`="Std. Error",
`type`="number",
`format`="zto"),
list(
`name`="z_value",
`title`="z-value",
`type`="number",
`format`="zto"),
list(
`name`="p_value",
`title`="p-value",
`type`="number",
`format`="zto,pvalue"),
list(
`name`="odds_ratio",
`title`="Odds Ratio",
`type`="number",
`format`="zto"),
list(
`name`="ci_lower",
`title`="OR 95% CI Lower",
`type`="number",
`format`="zto"),
list(
`name`="ci_upper",
`title`="OR 95% CI Upper",
`type`="number",
`format`="zto")),
clearWith=list(
"customPredictors",
"outcome",
"outcomePositive")))
self$add(jmvcore::Table$new(
options=options,
name="modelComparisonTable",
title="Model Performance Comparison",
visible="(compareModels && showPerformanceMetrics)",
columns=list(
list(
`name`="model",
`title`="Model",
`type`="text"),
list(
`name`="n_predictors",
`title`="Variables",
`type`="integer"),
list(
`name`="auc_training",
`title`="AUC (Training)",
`type`="number",
`format`="zto"),
list(
`name`="auc_validation",
`title`="AUC (Validation)",
`type`="number",
`format`="zto"),
list(
`name`="calibration_slope",
`title`="Calibration Slope",
`type`="number",
`format`="zto"),
list(
`name`="calibration_intercept",
`title`="Calibration Intercept",
`type`="number",
`format`="zto"),
list(
`name`="brier_score",
`title`="Brier Score",
`type`="number",
`format`="zto"),
list(
`name`="log_likelihood",
`title`="Log-Likelihood",
`type`="number",
`format`="zto"),
list(
`name`="aic",
`title`="AIC",
`type`="number",
`format`="zto"),
list(
`name`="bic",
`title`="BIC",
`type`="number",
`format`="zto")),
clearWith=list(
"outcome",
"outcomePositive",
"splitData")))
self$add(jmvcore::Table$new(
options=options,
name="validationResults",
title="Cross-Validation Results",
visible="(crossValidation)",
columns=list(
list(
`name`="model",
`title`="Model",
`type`="text"),
list(
`name`="cv_auc_mean",
`title`="CV AUC (Mean)",
`type`="number",
`format`="zto"),
list(
`name`="cv_auc_sd",
`title`="CV AUC (SD)",
`type`="number",
`format`="zto"),
list(
`name`="cv_calibration_slope",
`title`="CV Calibration Slope",
`type`="number",
`format`="zto"),
list(
`name`="cv_brier_score",
`title`="CV Brier Score",
`type`="number",
`format`="zto"),
list(
`name`="optimism",
`title`="Optimism",
`type`="number",
`format`="zto")),
clearWith=list(
"cvFolds",
"outcome",
"outcomePositive")))
self$add(jmvcore::Table$new(
options=options,
name="nriTable",
title="Net Reclassification Index",
visible="(calculateNRI)",
columns=list(
list(
`name`="comparison",
`title`="Comparison",
`type`="text"),
list(
`name`="nri_events",
`title`="NRI (Events)",
`type`="number",
`format`="zto"),
list(
`name`="nri_non_events",
`title`="NRI (Non-Events)",
`type`="number",
`format`="zto"),
list(
`name`="nri_overall",
`title`="NRI (Overall)",
`type`="number",
`format`="zto"),
list(
`name`="nri_p_value",
`title`="P-value",
`type`="number",
`format`="zto,pvalue")),
clearWith=list(
"nriThresholds",
"outcome")))
self$add(jmvcore::Table$new(
options=options,
name="idiTable",
title="Integrated Discrimination Index",
visible="(calculateIDI)",
columns=list(
list(
`name`="comparison",
`title`="Comparison",
`type`="text"),
list(
`name`="idi",
`title`="IDI",
`type`="number",
`format`="zto"),
list(
`name`="idi_p_value",
`title`="P-value",
`type`="number",
`format`="zto,pvalue"),
list(
`name`="discrimination_slope_new",
`title`="Disc. Slope (New)",
`type`="number",
`format`="zto"),
list(
`name`="discrimination_slope_old",
`title`="Disc. Slope (Reference)",
`type`="number",
`format`="zto")),
clearWith=list(
"outcome")))
self$add(jmvcore::Table$new(
options=options,
name="riskScoreTable",
title="Clinical Risk Score",
visible="(generateRiskScore)",
columns=list(
list(
`name`="variable",
`title`="Risk Factor",
`type`="text"),
list(
`name`="category",
`title`="Category",
`type`="text"),
list(
`name`="points",
`title`="Points",
`type`="integer"),
list(
`name`="coefficient",
`title`="Coefficient",
`type`="number",
`format`="zto"),
list(
`name`="interpretation",
`title`="Clinical Impact",
`type`="text")),
clearWith=list(
"riskScorePoints",
"outcome")))
self$add(jmvcore::Image$new(
options=options,
name="rocCurvesPlot",
title="ROC Curves Comparison",
width=600,
height=450,
renderFun=".plotROCCurves",
visible="(showROCCurves)",
clearWith=list(
"outcome",
"outcomePositive",
"splitData")))
self$add(jmvcore::Array$new(
options=options,
name="calibrationPlotsArray",
title="Calibration Plots",
visible="(showCalibrationPlots)",
template=jmvcore::Image$new(
options=options,
title="$key",
width=500,
height=400,
renderFun=".plotCalibration",
clearWith=list(
"outcome",
"outcomePositive",
"splitData"))))
self$add(jmvcore::Image$new(
options=options,
name="modelComparisonPlot",
title="Model Performance Comparison",
width=700,
height=500,
renderFun=".plotModelComparison",
visible="(compareModels)",
clearWith=list(
"outcome",
"outcomePositive")))
self$add(jmvcore::Image$new(
options=options,
name="validationPlot",
title="Cross-Validation Results",
width=600,
height=450,
renderFun=".plotValidation",
visible="(crossValidation || bootstrapValidation)",
clearWith=list(
"cvFolds",
"bootstrapReps")))
self$add(jmvcore::Html$new(
options=options,
name="dcaPreparationSummary",
title="Decision Curve Analysis Preparation",
visible="(exportForDCA)"))}))
modelbuilderBase <- if (requireNamespace("jmvcore", quietly=TRUE)) R6::R6Class(
"modelbuilderBase",
inherit = jmvcore::Analysis,
public = list(
initialize = function(options, data=NULL, datasetId="", analysisId="", revision=0) {
super$initialize(
package = "ClinicoPath",
name = "modelbuilder",
version = c(0,0,3),
options = options,
results = modelbuilderResults$new(options=options),
data = data,
datasetId = datasetId,
analysisId = analysisId,
revision = revision,
pause = NULL,
completeWhenFilled = FALSE,
requiresMissings = FALSE,
weightsSupport = 'auto')
}))
#' Prediction Model Builder
#'
#' Build and validate prediction models for medical decision making. Creates
#' multiple logistic regression models with predicted probabilities that can
#' be directly used in Decision Curve Analysis.
#'
#'
#' @examples
#' \donttest{
#' # example will be added
#'}
#' @param data The data as a data frame.
#' @param outcome Binary outcome variable to predict (e.g., disease presence,
#' adverse event occurrence).
#' @param outcomePositive Which level of the outcome variable represents the
#' positive case.
#' @param splitData Split data into training (70\%) and validation (30\%) sets
#' for unbiased model evaluation.
#' @param randomSeed Random seed for reproducible data splitting and results.
#' @param buildBasicModel Build a basic clinical model using demographic and
#' primary risk factors.
#' @param basicPredictors Variables for the basic clinical model (e.g., age,
#' sex, primary risk factors).
#' @param basicModelName Name for the basic model (used as column name for
#' predictions).
#' @param buildEnhancedModel Build an enhanced model with additional clinical
#' variables.
#' @param enhancedPredictors Variables for the enhanced model (should include
#' basic predictors plus additional ones).
#' @param enhancedModelName Name for the enhanced model (used as column name
#' for predictions).
#' @param buildBiomarkerModel Build a model incorporating biomarkers or
#' advanced diagnostics.
#' @param biomarkerPredictors Variables for the biomarker model (clinical
#' variables plus biomarkers).
#' @param biomarkerModelName Name for the biomarker model (used as column name
#' for predictions).
#' @param buildCustomModel Build a custom model with user-specified variables.
#' @param customPredictors Variables for the custom model.
#' @param customModelName Name for the custom model (used as column name for
#' predictions).
#' @param includeInteractions Include two-way interactions between predictor
#' variables.
#' @param interactionTerms Specify interactions (e.g., "age*sex,
#' diabetes*smoking"). Leave empty for all pairwise.
#' @param useStepwise Use stepwise variable selection to optimize models.
#' @param stepwiseDirection Direction for stepwise variable selection.
#' @param selectionCriterion Criterion for stepwise variable selection.
#' @param transformVariables Apply automatic transformations to continuous
#' variables.
#' @param transformMethod Type of transformation to apply to continuous
#' variables.
#' @param missingDataMethod Method for handling missing data in predictor
#' variables.
#' @param imputationSets Number of imputation sets for multiple imputation.
#' @param crossValidation Perform k-fold cross-validation for robust model
#' assessment.
#' @param cvFolds Number of folds for cross-validation.
#' @param bootstrapValidation Perform bootstrap validation to assess optimism
#' and shrinkage.
#' @param bootstrapReps Number of bootstrap replications for validation.
#' @param showModelSummary Display regression coefficients and model
#' statistics.
#' @param showPerformanceMetrics Display AUC, calibration, and other
#' performance measures.
#' @param showCalibrationPlots Display calibration plots for each model.
#' @param showROCCurves Display ROC curves for discrimination assessment.
#' @param compareModels Display side-by-side comparison of all models.
#' @param createPredictions Add predicted probability columns to dataset for
#' DCA use.
#' @param exportForDCA Format output specifically for use in Decision Curve
#' Analysis module.
#' @param calculateNRI Calculate NRI comparing enhanced models to basic model.
#' @param nriThresholds Risk thresholds for NRI calculation (comma-separated).
#' @param calculateIDI Calculate IDI comparing models.
#' @param penalizedRegression Use penalized regression (LASSO/Ridge) for
#' variable selection.
#' @param penaltyType Type of penalty for regularized regression.
#' @param generateRiskScore Create integer risk score from best model for
#' clinical use.
#' @param riskScorePoints Method for creating clinical risk score.
#' @return A results object containing:
#' \tabular{llllll}{
#' \code{results$instructions} \tab \tab \tab \tab \tab a html \cr
#' \code{results$dataSummary} \tab \tab \tab \tab \tab a html \cr
#' \code{results$dcaReadyMessage} \tab \tab \tab \tab \tab a html \cr
#' \code{results$basicModelSummary} \tab \tab \tab \tab \tab a table \cr
#' \code{results$enhancedModelSummary} \tab \tab \tab \tab \tab a table \cr
#' \code{results$biomarkerModelSummary} \tab \tab \tab \tab \tab a table \cr
#' \code{results$customModelSummary} \tab \tab \tab \tab \tab a table \cr
#' \code{results$modelComparisonTable} \tab \tab \tab \tab \tab a table \cr
#' \code{results$validationResults} \tab \tab \tab \tab \tab a table \cr
#' \code{results$nriTable} \tab \tab \tab \tab \tab a table \cr
#' \code{results$idiTable} \tab \tab \tab \tab \tab a table \cr
#' \code{results$riskScoreTable} \tab \tab \tab \tab \tab a table \cr
#' \code{results$rocCurvesPlot} \tab \tab \tab \tab \tab an image \cr
#' \code{results$calibrationPlotsArray} \tab \tab \tab \tab \tab an array of images \cr
#' \code{results$modelComparisonPlot} \tab \tab \tab \tab \tab an image \cr
#' \code{results$validationPlot} \tab \tab \tab \tab \tab an image \cr
#' \code{results$dcaPreparationSummary} \tab \tab \tab \tab \tab a html \cr
#' }
#'
#' Tables can be converted to data frames with \code{asDF} or \code{\link{as.data.frame}}. For example:
#'
#' \code{results$basicModelSummary$asDF}
#'
#' \code{as.data.frame(results$basicModelSummary)}
#'
#' @export
modelbuilder <- function(
data,
outcome,
outcomePositive,
splitData = TRUE,
randomSeed = 123,
buildBasicModel = TRUE,
basicPredictors,
basicModelName = "basic_model",
buildEnhancedModel = FALSE,
enhancedPredictors,
enhancedModelName = "enhanced_model",
buildBiomarkerModel = FALSE,
biomarkerPredictors,
biomarkerModelName = "biomarker_model",
buildCustomModel = FALSE,
customPredictors,
customModelName = "custom_model",
includeInteractions = FALSE,
interactionTerms = "",
useStepwise = FALSE,
stepwiseDirection = "both",
selectionCriterion = "aic",
transformVariables = FALSE,
transformMethod = "log",
missingDataMethod = "complete_cases",
imputationSets = 5,
crossValidation = FALSE,
cvFolds = 5,
bootstrapValidation = FALSE,
bootstrapReps = 1000,
showModelSummary = TRUE,
showPerformanceMetrics = TRUE,
showCalibrationPlots = TRUE,
showROCCurves = TRUE,
compareModels = TRUE,
createPredictions = TRUE,
exportForDCA = TRUE,
calculateNRI = FALSE,
nriThresholds = "0.05, 0.10, 0.20",
calculateIDI = FALSE,
penalizedRegression = FALSE,
penaltyType = "lasso",
generateRiskScore = FALSE,
riskScorePoints = "simple") {
if ( ! requireNamespace("jmvcore", quietly=TRUE))
stop("modelbuilder requires jmvcore to be installed (restart may be required)")
if ( ! missing(outcome)) outcome <- jmvcore::resolveQuo(jmvcore::enquo(outcome))
if ( ! missing(basicPredictors)) basicPredictors <- jmvcore::resolveQuo(jmvcore::enquo(basicPredictors))
if ( ! missing(enhancedPredictors)) enhancedPredictors <- jmvcore::resolveQuo(jmvcore::enquo(enhancedPredictors))
if ( ! missing(biomarkerPredictors)) biomarkerPredictors <- jmvcore::resolveQuo(jmvcore::enquo(biomarkerPredictors))
if ( ! missing(customPredictors)) customPredictors <- jmvcore::resolveQuo(jmvcore::enquo(customPredictors))
if (missing(data))
data <- jmvcore::marshalData(
parent.frame(),
`if`( ! missing(outcome), outcome, NULL),
`if`( ! missing(basicPredictors), basicPredictors, NULL),
`if`( ! missing(enhancedPredictors), enhancedPredictors, NULL),
`if`( ! missing(biomarkerPredictors), biomarkerPredictors, NULL),
`if`( ! missing(customPredictors), customPredictors, NULL))
for (v in outcome) if (v %in% names(data)) data[[v]] <- as.factor(data[[v]])
options <- modelbuilderOptions$new(
outcome = outcome,
outcomePositive = outcomePositive,
splitData = splitData,
randomSeed = randomSeed,
buildBasicModel = buildBasicModel,
basicPredictors = basicPredictors,
basicModelName = basicModelName,
buildEnhancedModel = buildEnhancedModel,
enhancedPredictors = enhancedPredictors,
enhancedModelName = enhancedModelName,
buildBiomarkerModel = buildBiomarkerModel,
biomarkerPredictors = biomarkerPredictors,
biomarkerModelName = biomarkerModelName,
buildCustomModel = buildCustomModel,
customPredictors = customPredictors,
customModelName = customModelName,
includeInteractions = includeInteractions,
interactionTerms = interactionTerms,
useStepwise = useStepwise,
stepwiseDirection = stepwiseDirection,
selectionCriterion = selectionCriterion,
transformVariables = transformVariables,
transformMethod = transformMethod,
missingDataMethod = missingDataMethod,
imputationSets = imputationSets,
crossValidation = crossValidation,
cvFolds = cvFolds,
bootstrapValidation = bootstrapValidation,
bootstrapReps = bootstrapReps,
showModelSummary = showModelSummary,
showPerformanceMetrics = showPerformanceMetrics,
showCalibrationPlots = showCalibrationPlots,
showROCCurves = showROCCurves,
compareModels = compareModels,
createPredictions = createPredictions,
exportForDCA = exportForDCA,
calculateNRI = calculateNRI,
nriThresholds = nriThresholds,
calculateIDI = calculateIDI,
penalizedRegression = penalizedRegression,
penaltyType = penaltyType,
generateRiskScore = generateRiskScore,
riskScorePoints = riskScorePoints)
analysis <- modelbuilderClass$new(
options = options,
data = data)
analysis$run()
analysis$results
}
sbalci/ClinicoPathJamoviModule documentation built on June 13, 2025, 9:34 a.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Defines functions modelbuilder
# This file is automatically generated, you probably don't want to edit this
modelbuilderOptions <- if (requireNamespace("jmvcore", quietly=TRUE)) R6::R6Class(
"modelbuilderOptions",
inherit = jmvcore::Options,
public = list(
initialize = function(
outcome = NULL,
outcomePositive = NULL,
splitData = TRUE,
randomSeed = 123,
buildBasicModel = TRUE,
basicPredictors = NULL,
basicModelName = "basic_model",
buildEnhancedModel = FALSE,
enhancedPredictors = NULL,
enhancedModelName = "enhanced_model",
buildBiomarkerModel = FALSE,
biomarkerPredictors = NULL,
biomarkerModelName = "biomarker_model",
buildCustomModel = FALSE,
customPredictors = NULL,
customModelName = "custom_model",
includeInteractions = FALSE,
interactionTerms = "",
useStepwise = FALSE,
stepwiseDirection = "both",
selectionCriterion = "aic",
transformVariables = FALSE,
transformMethod = "log",
missingDataMethod = "complete_cases",
imputationSets = 5,
crossValidation = FALSE,
cvFolds = 5,
bootstrapValidation = FALSE,
bootstrapReps = 1000,
showModelSummary = TRUE,
showPerformanceMetrics = TRUE,
showCalibrationPlots = TRUE,
showROCCurves = TRUE,
compareModels = TRUE,
createPredictions = TRUE,
exportForDCA = TRUE,
calculateNRI = FALSE,
nriThresholds = "0.05, 0.10, 0.20",
calculateIDI = FALSE,
penalizedRegression = FALSE,
penaltyType = "lasso",
generateRiskScore = FALSE,
riskScorePoints = "simple", ...) {
super$initialize(
package="ClinicoPath",
name="modelbuilder",
requiresData=TRUE,
...)
private$..outcome <- jmvcore::OptionVariable$new(
"outcome",
outcome,
suggested=list(
"nominal"),
permitted=list(
"factor"))
private$..outcomePositive <- jmvcore::OptionLevel$new(
"outcomePositive",
outcomePositive,
variable="(outcome)")
private$..splitData <- jmvcore::OptionBool$new(
"splitData",
splitData,
default=TRUE)
private$..randomSeed <- jmvcore::OptionNumber$new(
"randomSeed",
randomSeed,
default=123,
min=1,
max=99999)
private$..buildBasicModel <- jmvcore::OptionBool$new(
"buildBasicModel",
buildBasicModel,
default=TRUE)
private$..basicPredictors <- jmvcore::OptionVariables$new(
"basicPredictors",
basicPredictors,
suggested=list(
"continuous",
"ordinal",
"nominal"),
permitted=list(
"numeric",
"factor"))
private$..basicModelName <- jmvcore::OptionString$new(
"basicModelName",
basicModelName,
default="basic_model")
private$..buildEnhancedModel <- jmvcore::OptionBool$new(
"buildEnhancedModel",
buildEnhancedModel,
default=FALSE)
private$..enhancedPredictors <- jmvcore::OptionVariables$new(
"enhancedPredictors",
enhancedPredictors,
suggested=list(
"continuous",
"ordinal",
"nominal"),
permitted=list(
"numeric",
"factor"))
private$..enhancedModelName <- jmvcore::OptionString$new(
"enhancedModelName",
enhancedModelName,
default="enhanced_model")
private$..buildBiomarkerModel <- jmvcore::OptionBool$new(
"buildBiomarkerModel",
buildBiomarkerModel,
default=FALSE)
private$..biomarkerPredictors <- jmvcore::OptionVariables$new(
"biomarkerPredictors",
biomarkerPredictors,
suggested=list(
"continuous",
"ordinal",
"nominal"),
permitted=list(
"numeric",
"factor"))
private$..biomarkerModelName <- jmvcore::OptionString$new(
"biomarkerModelName",
biomarkerModelName,
default="biomarker_model")
private$..buildCustomModel <- jmvcore::OptionBool$new(
"buildCustomModel",
buildCustomModel,
default=FALSE)
private$..customPredictors <- jmvcore::OptionVariables$new(
"customPredictors",
customPredictors,
suggested=list(
"continuous",
"ordinal",
"nominal"),
permitted=list(
"numeric",
"factor"))
private$..customModelName <- jmvcore::OptionString$new(
"customModelName",
customModelName,
default="custom_model")
private$..includeInteractions <- jmvcore::OptionBool$new(
"includeInteractions",
includeInteractions,
default=FALSE)
private$..interactionTerms <- jmvcore::OptionString$new(
"interactionTerms",
interactionTerms,
default="")
private$..useStepwise <- jmvcore::OptionBool$new(
"useStepwise",
useStepwise,
default=FALSE)
private$..stepwiseDirection <- jmvcore::OptionList$new(
"stepwiseDirection",
stepwiseDirection,
options=list(
"both",
"forward",
"backward"),
default="both")
private$..selectionCriterion <- jmvcore::OptionList$new(
"selectionCriterion",
selectionCriterion,
options=list(
"aic",
"bic"),
default="aic")
private$..transformVariables <- jmvcore::OptionBool$new(
"transformVariables",
transformVariables,
default=FALSE)
private$..transformMethod <- jmvcore::OptionList$new(
"transformMethod",
transformMethod,
options=list(
"log",
"polynomial",
"spline"),
default="log")
private$..missingDataMethod <- jmvcore::OptionList$new(
"missingDataMethod",
missingDataMethod,
options=list(
"complete_cases",
"mean_imputation",
"multiple_imputation",
"exclude_missing"),
default="complete_cases")
private$..imputationSets <- jmvcore::OptionNumber$new(
"imputationSets",
imputationSets,
default=5,
min=3,
max=20)
private$..crossValidation <- jmvcore::OptionBool$new(
"crossValidation",
crossValidation,
default=FALSE)
private$..cvFolds <- jmvcore::OptionNumber$new(
"cvFolds",
cvFolds,
default=5,
min=3,
max=10)
private$..bootstrapValidation <- jmvcore::OptionBool$new(
"bootstrapValidation",
bootstrapValidation,
default=FALSE)
private$..bootstrapReps <- jmvcore::OptionNumber$new(
"bootstrapReps",
bootstrapReps,
default=1000,
min=100,
max=5000)
private$..showModelSummary <- jmvcore::OptionBool$new(
"showModelSummary",
showModelSummary,
default=TRUE)
private$..showPerformanceMetrics <- jmvcore::OptionBool$new(
"showPerformanceMetrics",
showPerformanceMetrics,
default=TRUE)
private$..showCalibrationPlots <- jmvcore::OptionBool$new(
"showCalibrationPlots",
showCalibrationPlots,
default=TRUE)
private$..showROCCurves <- jmvcore::OptionBool$new(
"showROCCurves",
showROCCurves,
default=TRUE)
private$..compareModels <- jmvcore::OptionBool$new(
"compareModels",
compareModels,
default=TRUE)
private$..createPredictions <- jmvcore::OptionBool$new(
"createPredictions",
createPredictions,
default=TRUE)
private$..exportForDCA <- jmvcore::OptionBool$new(
"exportForDCA",
exportForDCA,
default=TRUE)
private$..calculateNRI <- jmvcore::OptionBool$new(
"calculateNRI",
calculateNRI,
default=FALSE)
private$..nriThresholds <- jmvcore::OptionString$new(
"nriThresholds",
nriThresholds,
default="0.05, 0.10, 0.20")
private$..calculateIDI <- jmvcore::OptionBool$new(
"calculateIDI",
calculateIDI,
default=FALSE)
private$..penalizedRegression <- jmvcore::OptionBool$new(
"penalizedRegression",
penalizedRegression,
default=FALSE)
private$..penaltyType <- jmvcore::OptionList$new(
"penaltyType",
penaltyType,
options=list(
"lasso",
"ridge",
"elastic_net"),
default="lasso")
private$..generateRiskScore <- jmvcore::OptionBool$new(
"generateRiskScore",
generateRiskScore,
default=FALSE)
private$..riskScorePoints <- jmvcore::OptionList$new(
"riskScorePoints",
riskScorePoints,
options=list(
"framingham",
"simple",
"deciles"),
default="simple")
self$.addOption(private$..outcome)
self$.addOption(private$..outcomePositive)
self$.addOption(private$..splitData)
self$.addOption(private$..randomSeed)
self$.addOption(private$..buildBasicModel)
self$.addOption(private$..basicPredictors)
self$.addOption(private$..basicModelName)
self$.addOption(private$..buildEnhancedModel)
self$.addOption(private$..enhancedPredictors)
self$.addOption(private$..enhancedModelName)
self$.addOption(private$..buildBiomarkerModel)
self$.addOption(private$..biomarkerPredictors)
self$.addOption(private$..biomarkerModelName)
self$.addOption(private$..buildCustomModel)
self$.addOption(private$..customPredictors)
self$.addOption(private$..customModelName)
self$.addOption(private$..includeInteractions)
self$.addOption(private$..interactionTerms)
self$.addOption(private$..useStepwise)
self$.addOption(private$..stepwiseDirection)
self$.addOption(private$..selectionCriterion)
self$.addOption(private$..transformVariables)
self$.addOption(private$..transformMethod)
self$.addOption(private$..missingDataMethod)
self$.addOption(private$..imputationSets)
self$.addOption(private$..crossValidation)
self$.addOption(private$..cvFolds)
self$.addOption(private$..bootstrapValidation)
self$.addOption(private$..bootstrapReps)
self$.addOption(private$..showModelSummary)
self$.addOption(private$..showPerformanceMetrics)
self$.addOption(private$..showCalibrationPlots)
self$.addOption(private$..showROCCurves)
self$.addOption(private$..compareModels)
self$.addOption(private$..createPredictions)
self$.addOption(private$..exportForDCA)
self$.addOption(private$..calculateNRI)
self$.addOption(private$..nriThresholds)
self$.addOption(private$..calculateIDI)
self$.addOption(private$..penalizedRegression)
self$.addOption(private$..penaltyType)
self$.addOption(private$..generateRiskScore)
self$.addOption(private$..riskScorePoints)
}),
active = list(
outcome = function() private$..outcome$value,
outcomePositive = function() private$..outcomePositive$value,
splitData = function() private$..splitData$value,
randomSeed = function() private$..randomSeed$value,
buildBasicModel = function() private$..buildBasicModel$value,
basicPredictors = function() private$..basicPredictors$value,
basicModelName = function() private$..basicModelName$value,
buildEnhancedModel = function() private$..buildEnhancedModel$value,
enhancedPredictors = function() private$..enhancedPredictors$value,
enhancedModelName = function() private$..enhancedModelName$value,
buildBiomarkerModel = function() private$..buildBiomarkerModel$value,
biomarkerPredictors = function() private$..biomarkerPredictors$value,
biomarkerModelName = function() private$..biomarkerModelName$value,
buildCustomModel = function() private$..buildCustomModel$value,
customPredictors = function() private$..customPredictors$value,
customModelName = function() private$..customModelName$value,
includeInteractions = function() private$..includeInteractions$value,
interactionTerms = function() private$..interactionTerms$value,
useStepwise = function() private$..useStepwise$value,
stepwiseDirection = function() private$..stepwiseDirection$value,
selectionCriterion = function() private$..selectionCriterion$value,
transformVariables = function() private$..transformVariables$value,
transformMethod = function() private$..transformMethod$value,
missingDataMethod = function() private$..missingDataMethod$value,
imputationSets = function() private$..imputationSets$value,
crossValidation = function() private$..crossValidation$value,
cvFolds = function() private$..cvFolds$value,
bootstrapValidation = function() private$..bootstrapValidation$value,
bootstrapReps = function() private$..bootstrapReps$value,
showModelSummary = function() private$..showModelSummary$value,
showPerformanceMetrics = function() private$..showPerformanceMetrics$value,
showCalibrationPlots = function() private$..showCalibrationPlots$value,
showROCCurves = function() private$..showROCCurves$value,
compareModels = function() private$..compareModels$value,
createPredictions = function() private$..createPredictions$value,
exportForDCA = function() private$..exportForDCA$value,
calculateNRI = function() private$..calculateNRI$value,
nriThresholds = function() private$..nriThresholds$value,
calculateIDI = function() private$..calculateIDI$value,
penalizedRegression = function() private$..penalizedRegression$value,
penaltyType = function() private$..penaltyType$value,
generateRiskScore = function() private$..generateRiskScore$value,
riskScorePoints = function() private$..riskScorePoints$value),
private = list(
..outcome = NA,
..outcomePositive = NA,
..splitData = NA,
..randomSeed = NA,
..buildBasicModel = NA,
..basicPredictors = NA,
..basicModelName = NA,
..buildEnhancedModel = NA,
..enhancedPredictors = NA,
..enhancedModelName = NA,
..buildBiomarkerModel = NA,
..biomarkerPredictors = NA,
..biomarkerModelName = NA,
..buildCustomModel = NA,
..customPredictors = NA,
..customModelName = NA,
..includeInteractions = NA,
..interactionTerms = NA,
..useStepwise = NA,
..stepwiseDirection = NA,
..selectionCriterion = NA,
..transformVariables = NA,
..transformMethod = NA,
..missingDataMethod = NA,
..imputationSets = NA,
..crossValidation = NA,
..cvFolds = NA,
..bootstrapValidation = NA,
..bootstrapReps = NA,
..showModelSummary = NA,
..showPerformanceMetrics = NA,
..showCalibrationPlots = NA,
..showROCCurves = NA,
..compareModels = NA,
..createPredictions = NA,
..exportForDCA = NA,
..calculateNRI = NA,
..nriThresholds = NA,
..calculateIDI = NA,
..penalizedRegression = NA,
..penaltyType = NA,
..generateRiskScore = NA,
..riskScorePoints = NA)
)
modelbuilderResults <- if (requireNamespace("jmvcore", quietly=TRUE)) R6::R6Class(
"modelbuilderResults",
inherit = jmvcore::Group,
active = list(
instructions = function() private$.items[["instructions"]],
dataSummary = function() private$.items[["dataSummary"]],
dcaReadyMessage = function() private$.items[["dcaReadyMessage"]],
basicModelSummary = function() private$.items[["basicModelSummary"]],
enhancedModelSummary = function() private$.items[["enhancedModelSummary"]],
biomarkerModelSummary = function() private$.items[["biomarkerModelSummary"]],
customModelSummary = function() private$.items[["customModelSummary"]],
modelComparisonTable = function() private$.items[["modelComparisonTable"]],
validationResults = function() private$.items[["validationResults"]],
nriTable = function() private$.items[["nriTable"]],
idiTable = function() private$.items[["idiTable"]],
riskScoreTable = function() private$.items[["riskScoreTable"]],
rocCurvesPlot = function() private$.items[["rocCurvesPlot"]],
calibrationPlotsArray = function() private$.items[["calibrationPlotsArray"]],
modelComparisonPlot = function() private$.items[["modelComparisonPlot"]],
validationPlot = function() private$.items[["validationPlot"]],
dcaPreparationSummary = function() private$.items[["dcaPreparationSummary"]]),
private = list(),
public=list(
initialize=function(options) {
super$initialize(
options=options,
name="",
title="Prediction Model Builder",
refs=list(
"glmnet",
"rms",
"pROC",
"mice",
"VIM",
"ClinicoPathJamoviModule"))
self$add(jmvcore::Html$new(
options=options,
name="instructions",
title="Instructions",
visible=TRUE))
self$add(jmvcore::Html$new(
options=options,
name="dataSummary",
title="Data Summary",
visible=TRUE))
self$add(jmvcore::Html$new(
options=options,
name="dcaReadyMessage",
title="Ready for Decision Curve Analysis",
visible="(createPredictions)"))
self$add(jmvcore::Table$new(
options=options,
name="basicModelSummary",
title="Basic Clinical Model",
visible="(buildBasicModel && showModelSummary)",
columns=list(
list(
`name`="term",
`title`="Variable",
`type`="text"),
list(
`name`="estimate",
`title`="Coefficient",
`type`="number",
`format`="zto"),
list(
`name`="std_error",
`title`="Std. Error",
`type`="number",
`format`="zto"),
list(
`name`="z_value",
`title`="z-value",
`type`="number",
`format`="zto"),
list(
`name`="p_value",
`title`="p-value",
`type`="number",
`format`="zto,pvalue"),
list(
`name`="odds_ratio",
`title`="Odds Ratio",
`type`="number",
`format`="zto"),
list(
`name`="ci_lower",
`title`="OR 95% CI Lower",
`type`="number",
`format`="zto"),
list(
`name`="ci_upper",
`title`="OR 95% CI Upper",
`type`="number",
`format`="zto")),
clearWith=list(
"basicPredictors",
"outcome",
"outcomePositive")))
self$add(jmvcore::Table$new(
options=options,
name="enhancedModelSummary",
title="Enhanced Clinical Model",
visible="(buildEnhancedModel && showModelSummary)",
columns=list(
list(
`name`="term",
`title`="Variable",
`type`="text"),
list(
`name`="estimate",
`title`="Coefficient",
`type`="number",
`format`="zto"),
list(
`name`="std_error",
`title`="Std. Error",
`type`="number",
`format`="zto"),
list(
`name`="z_value",
`title`="z-value",
`type`="number",
`format`="zto"),
list(
`name`="p_value",
`title`="p-value",
`type`="number",
`format`="zto,pvalue"),
list(
`name`="odds_ratio",
`title`="Odds Ratio",
`type`="number",
`format`="zto"),
list(
`name`="ci_lower",
`title`="OR 95% CI Lower",
`type`="number",
`format`="zto"),
list(
`name`="ci_upper",
`title`="OR 95% CI Upper",
`type`="number",
`format`="zto")),
clearWith=list(
"enhancedPredictors",
"outcome",
"outcomePositive")))
self$add(jmvcore::Table$new(
options=options,
name="biomarkerModelSummary",
title="Biomarker Model",
visible="(buildBiomarkerModel && showModelSummary)",
columns=list(
list(
`name`="term",
`title`="Variable",
`type`="text"),
list(
`name`="estimate",
`title`="Coefficient",
`type`="number",
`format`="zto"),
list(
`name`="std_error",
`title`="Std. Error",
`type`="number",
`format`="zto"),
list(
`name`="z_value",
`title`="z-value",
`type`="number",
`format`="zto"),
list(
`name`="p_value",
`title`="p-value",
`type`="number",
`format`="zto,pvalue"),
list(
`name`="odds_ratio",
`title`="Odds Ratio",
`type`="number",
`format`="zto"),
list(
`name`="ci_lower",
`title`="OR 95% CI Lower",
`type`="number",
`format`="zto"),
list(
`name`="ci_upper",
`title`="OR 95% CI Upper",
`type`="number",
`format`="zto")),
clearWith=list(
"biomarkerPredictors",
"outcome",
"outcomePositive")))
self$add(jmvcore::Table$new(
options=options,
name="customModelSummary",
title="Custom Model",
visible="(buildCustomModel && showModelSummary)",
columns=list(
list(
`name`="term",
`title`="Variable",
`type`="text"),
list(
`name`="estimate",
`title`="Coefficient",
`type`="number",
`format`="zto"),
list(
`name`="std_error",
`title`="Std. Error",
`type`="number",
`format`="zto"),
list(
`name`="z_value",
`title`="z-value",
`type`="number",
`format`="zto"),
list(
`name`="p_value",
`title`="p-value",
`type`="number",
`format`="zto,pvalue"),
list(
`name`="odds_ratio",
`title`="Odds Ratio",
`type`="number",
`format`="zto"),
list(
`name`="ci_lower",
`title`="OR 95% CI Lower",
`type`="number",
`format`="zto"),
list(
`name`="ci_upper",
`title`="OR 95% CI Upper",
`type`="number",
`format`="zto")),
clearWith=list(
"customPredictors",
"outcome",
"outcomePositive")))
self$add(jmvcore::Table$new(
options=options,
name="modelComparisonTable",
title="Model Performance Comparison",
visible="(compareModels && showPerformanceMetrics)",
columns=list(
list(
`name`="model",
`title`="Model",
`type`="text"),
list(
`name`="n_predictors",
`title`="Variables",
`type`="integer"),
list(
`name`="auc_training",
`title`="AUC (Training)",
`type`="number",
`format`="zto"),
list(
`name`="auc_validation",
`title`="AUC (Validation)",
`type`="number",
`format`="zto"),
list(
`name`="calibration_slope",
`title`="Calibration Slope",
`type`="number",
`format`="zto"),
list(
`name`="calibration_intercept",
`title`="Calibration Intercept",
`type`="number",
`format`="zto"),
list(
`name`="brier_score",
`title`="Brier Score",
`type`="number",
`format`="zto"),
list(
`name`="log_likelihood",
`title`="Log-Likelihood",
`type`="number",
`format`="zto"),
list(
`name`="aic",
`title`="AIC",
`type`="number",
`format`="zto"),
list(
`name`="bic",
`title`="BIC",
`type`="number",
`format`="zto")),
clearWith=list(
"outcome",
"outcomePositive",
"splitData")))
self$add(jmvcore::Table$new(
options=options,
name="validationResults",
title="Cross-Validation Results",
visible="(crossValidation)",
columns=list(
list(
`name`="model",
`title`="Model",
`type`="text"),
list(
`name`="cv_auc_mean",
`title`="CV AUC (Mean)",
`type`="number",
`format`="zto"),
list(
`name`="cv_auc_sd",
`title`="CV AUC (SD)",
`type`="number",
`format`="zto"),
list(
`name`="cv_calibration_slope",
`title`="CV Calibration Slope",
`type`="number",
`format`="zto"),
list(
`name`="cv_brier_score",
`title`="CV Brier Score",
`type`="number",
`format`="zto"),
list(
`name`="optimism",
`title`="Optimism",
`type`="number",
`format`="zto")),
clearWith=list(
"cvFolds",
"outcome",
"outcomePositive")))
self$add(jmvcore::Table$new(
options=options,
name="nriTable",
title="Net Reclassification Index",
visible="(calculateNRI)",
columns=list(
list(
`name`="comparison",
`title`="Comparison",
`type`="text"),
list(
`name`="nri_events",
`title`="NRI (Events)",
`type`="number",
`format`="zto"),
list(
`name`="nri_non_events",
`title`="NRI (Non-Events)",
`type`="number",
`format`="zto"),
list(
`name`="nri_overall",
`title`="NRI (Overall)",
`type`="number",
`format`="zto"),
list(
`name`="nri_p_value",
`title`="P-value",
`type`="number",
`format`="zto,pvalue")),
clearWith=list(
"nriThresholds",
"outcome")))
self$add(jmvcore::Table$new(
options=options,
name="idiTable",
title="Integrated Discrimination Index",
visible="(calculateIDI)",
columns=list(
list(
`name`="comparison",
`title`="Comparison",
`type`="text"),
list(
`name`="idi",
`title`="IDI",
`type`="number",
`format`="zto"),
list(
`name`="idi_p_value",
`title`="P-value",
`type`="number",
`format`="zto,pvalue"),
list(
`name`="discrimination_slope_new",
`title`="Disc. Slope (New)",
`type`="number",
`format`="zto"),
list(
`name`="discrimination_slope_old",
`title`="Disc. Slope (Reference)",
`type`="number",
`format`="zto")),
clearWith=list(
"outcome")))
self$add(jmvcore::Table$new(
options=options,
name="riskScoreTable",
title="Clinical Risk Score",
visible="(generateRiskScore)",
columns=list(
list(
`name`="variable",
`title`="Risk Factor",
`type`="text"),
list(
`name`="category",
`title`="Category",
`type`="text"),
list(
`name`="points",
`title`="Points",
`type`="integer"),
list(
`name`="coefficient",
`title`="Coefficient",
`type`="number",
`format`="zto"),
list(
`name`="interpretation",
`title`="Clinical Impact",
`type`="text")),
clearWith=list(
"riskScorePoints",
"outcome")))
self$add(jmvcore::Image$new(
options=options,
name="rocCurvesPlot",
title="ROC Curves Comparison",
width=600,
height=450,
renderFun=".plotROCCurves",
visible="(showROCCurves)",
clearWith=list(
"outcome",
"outcomePositive",
"splitData")))
self$add(jmvcore::Array$new(
options=options,
name="calibrationPlotsArray",
title="Calibration Plots",
visible="(showCalibrationPlots)",
template=jmvcore::Image$new(
options=options,
title="$key",
width=500,
height=400,
renderFun=".plotCalibration",
clearWith=list(
"outcome",
"outcomePositive",
"splitData"))))
self$add(jmvcore::Image$new(
options=options,
name="modelComparisonPlot",
title="Model Performance Comparison",
width=700,
height=500,
renderFun=".plotModelComparison",
visible="(compareModels)",
clearWith=list(
"outcome",
"outcomePositive")))
self$add(jmvcore::Image$new(
options=options,
name="validationPlot",
title="Cross-Validation Results",
width=600,
height=450,
renderFun=".plotValidation",
visible="(crossValidation || bootstrapValidation)",
clearWith=list(
"cvFolds",
"bootstrapReps")))
self$add(jmvcore::Html$new(
options=options,
name="dcaPreparationSummary",
title="Decision Curve Analysis Preparation",
visible="(exportForDCA)"))}))
modelbuilderBase <- if (requireNamespace("jmvcore", quietly=TRUE)) R6::R6Class(
"modelbuilderBase",
inherit = jmvcore::Analysis,
public = list(
initialize = function(options, data=NULL, datasetId="", analysisId="", revision=0) {
super$initialize(
package = "ClinicoPath",
name = "modelbuilder",
version = c(0,0,3),
options = options,
results = modelbuilderResults$new(options=options),
data = data,
datasetId = datasetId,
analysisId = analysisId,
revision = revision,
pause = NULL,
completeWhenFilled = FALSE,
requiresMissings = FALSE,
weightsSupport = 'auto')
}))
#' Prediction Model Builder
#'
#' Build and validate prediction models for medical decision making. Creates
#' multiple logistic regression models with predicted probabilities that can
#' be directly used in Decision Curve Analysis.
#'
#'
#' @examples
#' \donttest{
#' # example will be added
#'}
#' @param data The data as a data frame.
#' @param outcome Binary outcome variable to predict (e.g., disease presence,
#' adverse event occurrence).
#' @param outcomePositive Which level of the outcome variable represents the
#' positive case.
#' @param splitData Split data into training (70\%) and validation (30\%) sets
#' for unbiased model evaluation.
#' @param randomSeed Random seed for reproducible data splitting and results.
#' @param buildBasicModel Build a basic clinical model using demographic and
#' primary risk factors.
#' @param basicPredictors Variables for the basic clinical model (e.g., age,
#' sex, primary risk factors).
#' @param basicModelName Name for the basic model (used as column name for
#' predictions).
#' @param buildEnhancedModel Build an enhanced model with additional clinical
#' variables.
#' @param enhancedPredictors Variables for the enhanced model (should include
#' basic predictors plus additional ones).
#' @param enhancedModelName Name for the enhanced model (used as column name
#' for predictions).
#' @param buildBiomarkerModel Build a model incorporating biomarkers or
#' advanced diagnostics.
#' @param biomarkerPredictors Variables for the biomarker model (clinical
#' variables plus biomarkers).
#' @param biomarkerModelName Name for the biomarker model (used as column name
#' for predictions).
#' @param buildCustomModel Build a custom model with user-specified variables.
#' @param customPredictors Variables for the custom model.
#' @param customModelName Name for the custom model (used as column name for
#' predictions).
#' @param includeInteractions Include two-way interactions between predictor
#' variables.
#' @param interactionTerms Specify interactions (e.g., "age*sex,
#' diabetes*smoking"). Leave empty for all pairwise.
#' @param useStepwise Use stepwise variable selection to optimize models.
#' @param stepwiseDirection Direction for stepwise variable selection.
#' @param selectionCriterion Criterion for stepwise variable selection.
#' @param transformVariables Apply automatic transformations to continuous
#' variables.
#' @param transformMethod Type of transformation to apply to continuous
#' variables.
#' @param missingDataMethod Method for handling missing data in predictor
#' variables.
#' @param imputationSets Number of imputation sets for multiple imputation.
#' @param crossValidation Perform k-fold cross-validation for robust model
#' assessment.
#' @param cvFolds Number of folds for cross-validation.
#' @param bootstrapValidation Perform bootstrap validation to assess optimism
#' and shrinkage.
#' @param bootstrapReps Number of bootstrap replications for validation.
#' @param showModelSummary Display regression coefficients and model
#' statistics.
#' @param showPerformanceMetrics Display AUC, calibration, and other
#' performance measures.
#' @param showCalibrationPlots Display calibration plots for each model.
#' @param showROCCurves Display ROC curves for discrimination assessment.
#' @param compareModels Display side-by-side comparison of all models.
#' @param createPredictions Add predicted probability columns to dataset for
#' DCA use.
#' @param exportForDCA Format output specifically for use in Decision Curve
#' Analysis module.
#' @param calculateNRI Calculate NRI comparing enhanced models to basic model.
#' @param nriThresholds Risk thresholds for NRI calculation (comma-separated).
#' @param calculateIDI Calculate IDI comparing models.
#' @param penalizedRegression Use penalized regression (LASSO/Ridge) for
#' variable selection.
#' @param penaltyType Type of penalty for regularized regression.
#' @param generateRiskScore Create integer risk score from best model for
#' clinical use.
#' @param riskScorePoints Method for creating clinical risk score.
#' @return A results object containing:
#' \tabular{llllll}{
#' \code{results$instructions} \tab \tab \tab \tab \tab a html \cr
#' \code{results$dataSummary} \tab \tab \tab \tab \tab a html \cr
#' \code{results$dcaReadyMessage} \tab \tab \tab \tab \tab a html \cr
#' \code{results$basicModelSummary} \tab \tab \tab \tab \tab a table \cr
#' \code{results$enhancedModelSummary} \tab \tab \tab \tab \tab a table \cr
#' \code{results$biomarkerModelSummary} \tab \tab \tab \tab \tab a table \cr
#' \code{results$customModelSummary} \tab \tab \tab \tab \tab a table \cr
#' \code{results$modelComparisonTable} \tab \tab \tab \tab \tab a table \cr
#' \code{results$validationResults} \tab \tab \tab \tab \tab a table \cr
#' \code{results$nriTable} \tab \tab \tab \tab \tab a table \cr
#' \code{results$idiTable} \tab \tab \tab \tab \tab a table \cr
#' \code{results$riskScoreTable} \tab \tab \tab \tab \tab a table \cr
#' \code{results$rocCurvesPlot} \tab \tab \tab \tab \tab an image \cr
#' \code{results$calibrationPlotsArray} \tab \tab \tab \tab \tab an array of images \cr
#' \code{results$modelComparisonPlot} \tab \tab \tab \tab \tab an image \cr
#' \code{results$validationPlot} \tab \tab \tab \tab \tab an image \cr
#' \code{results$dcaPreparationSummary} \tab \tab \tab \tab \tab a html \cr
#' }
#'
#' Tables can be converted to data frames with \code{asDF} or \code{\link{as.data.frame}}. For example:
#'
#' \code{results$basicModelSummary$asDF}
#'
#' \code{as.data.frame(results$basicModelSummary)}
#'
#' @export
modelbuilder <- function(
data,
outcome,
outcomePositive,
splitData = TRUE,
randomSeed = 123,
buildBasicModel = TRUE,
basicPredictors,
basicModelName = "basic_model",
buildEnhancedModel = FALSE,
enhancedPredictors,
enhancedModelName = "enhanced_model",
buildBiomarkerModel = FALSE,
biomarkerPredictors,
biomarkerModelName = "biomarker_model",
buildCustomModel = FALSE,
customPredictors,
customModelName = "custom_model",
includeInteractions = FALSE,
interactionTerms = "",
useStepwise = FALSE,
stepwiseDirection = "both",
selectionCriterion = "aic",
transformVariables = FALSE,
transformMethod = "log",
missingDataMethod = "complete_cases",
imputationSets = 5,
crossValidation = FALSE,
cvFolds = 5,
bootstrapValidation = FALSE,
bootstrapReps = 1000,
showModelSummary = TRUE,
showPerformanceMetrics = TRUE,
showCalibrationPlots = TRUE,
showROCCurves = TRUE,
compareModels = TRUE,
createPredictions = TRUE,
exportForDCA = TRUE,
calculateNRI = FALSE,
nriThresholds = "0.05, 0.10, 0.20",
calculateIDI = FALSE,
penalizedRegression = FALSE,
penaltyType = "lasso",
generateRiskScore = FALSE,
riskScorePoints = "simple") {
if ( ! requireNamespace("jmvcore", quietly=TRUE))
stop("modelbuilder requires jmvcore to be installed (restart may be required)")
if ( ! missing(outcome)) outcome <- jmvcore::resolveQuo(jmvcore::enquo(outcome))
if ( ! missing(basicPredictors)) basicPredictors <- jmvcore::resolveQuo(jmvcore::enquo(basicPredictors))
if ( ! missing(enhancedPredictors)) enhancedPredictors <- jmvcore::resolveQuo(jmvcore::enquo(enhancedPredictors))
if ( ! missing(biomarkerPredictors)) biomarkerPredictors <- jmvcore::resolveQuo(jmvcore::enquo(biomarkerPredictors))
if ( ! missing(customPredictors)) customPredictors <- jmvcore::resolveQuo(jmvcore::enquo(customPredictors))
if (missing(data))
data <- jmvcore::marshalData(
parent.frame(),
`if`( ! missing(outcome), outcome, NULL),
`if`( ! missing(basicPredictors), basicPredictors, NULL),
`if`( ! missing(enhancedPredictors), enhancedPredictors, NULL),
`if`( ! missing(biomarkerPredictors), biomarkerPredictors, NULL),
`if`( ! missing(customPredictors), customPredictors, NULL))
for (v in outcome) if (v %in% names(data)) data[[v]] <- as.factor(data[[v]])
options <- modelbuilderOptions$new(
outcome = outcome,
outcomePositive = outcomePositive,
splitData = splitData,
randomSeed = randomSeed,
buildBasicModel = buildBasicModel,
basicPredictors = basicPredictors,
basicModelName = basicModelName,
buildEnhancedModel = buildEnhancedModel,
enhancedPredictors = enhancedPredictors,
enhancedModelName = enhancedModelName,
buildBiomarkerModel = buildBiomarkerModel,
biomarkerPredictors = biomarkerPredictors,
biomarkerModelName = biomarkerModelName,
buildCustomModel = buildCustomModel,
customPredictors = customPredictors,
customModelName = customModelName,
includeInteractions = includeInteractions,
interactionTerms = interactionTerms,
useStepwise = useStepwise,
stepwiseDirection = stepwiseDirection,
selectionCriterion = selectionCriterion,
transformVariables = transformVariables,
transformMethod = transformMethod,
missingDataMethod = missingDataMethod,
imputationSets = imputationSets,
crossValidation = crossValidation,
cvFolds = cvFolds,
bootstrapValidation = bootstrapValidation,
bootstrapReps = bootstrapReps,
showModelSummary = showModelSummary,
showPerformanceMetrics = showPerformanceMetrics,
showCalibrationPlots = showCalibrationPlots,
showROCCurves = showROCCurves,
compareModels = compareModels,
createPredictions = createPredictions,
exportForDCA = exportForDCA,
calculateNRI = calculateNRI,
nriThresholds = nriThresholds,
calculateIDI = calculateIDI,
penalizedRegression = penalizedRegression,
penaltyType = penaltyType,
generateRiskScore = generateRiskScore,
riskScorePoints = riskScorePoints)
analysis <- modelbuilderClass$new(
options = options,
data = data)
analysis$run()
analysis$results
}
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.