Optimize diagnostic test panels by evaluating various combination strategies including parallel testing (cotest), sequential testing, and repeated tests. Creates decision trees to minimize cost while maximizing accuracy.
Usage
decisionpanel(
data,
tests,
testLevels = "",
gold,
goldPositive,
useCosts = FALSE,
testCosts = "",
fpCost = 100,
fnCost = 1000,
strategies = "all",
parallelRules = "any",
customThreshold = 2,
maxTests = 3,
sequentialStop = "positive",
optimizationCriteria = "accuracy",
minSensitivity = 0.8,
minSpecificity = 0.8,
createTree = FALSE,
treeMethod = "costSensitive",
maxDepth = 5,
minSplit = 20,
showAllCombinations = FALSE,
topN = 10,
compareStrategies = TRUE,
bootstrap = FALSE,
bootReps = 1000,
plotTree = TRUE,
plotComparison = TRUE,
plotCostEffect = TRUE,
plotROC = FALSE,
prevalence = 0,
crossValidate = FALSE,
nFolds = 5,
seed = 12345
)Arguments
- data
The data as a data frame.
- tests
Variables representing different diagnostic tests. Can select multiple tests (up to 10) for panel optimization.
- testLevels
String specifying positive test levels. Can be: - Single value (used for all tests): "Positive" - Multiple values (comma-separated): "Positive,Yes,1" - Empty string (auto-detection): ""
- gold
The gold standard variable for disease classification.
- goldPositive
The level of the gold standard variable that indicates disease presence.
- useCosts
Include cost considerations in the optimization process.
- testCosts
Comma-separated costs for each test in the same order as selected tests. Example: "10,25,50" for three tests.
- fpCost
Cost or harm associated with a false positive result.
- fnCost
Cost or harm associated with a false negative result.
- strategies
Which testing strategies to evaluate in the analysis.
- parallelRules
Rule for combining results in parallel testing.
- customThreshold
Number of positive tests required for overall positive result (when using custom rule).
- maxTests
Maximum number of tests to combine in any strategy.
- sequentialStop
When to stop testing in sequential strategies.
- optimizationCriteria
Primary criterion for optimizing test panels.
- minSensitivity
Minimum sensitivity constraint for panel selection.
- minSpecificity
Minimum specificity constraint for panel selection.
- createTree
Generate an optimal decision tree for test sequencing.
- treeMethod
Method for constructing the decision tree.
- maxDepth
Maximum depth of the decision tree.
- minSplit
Minimum number of observations required to split a node.
- showAllCombinations
Display performance metrics for all possible test combinations.
- topN
Number of best-performing panels to display in results.
- compareStrategies
Show comparative analysis of different testing strategies.
- bootstrap
Calculate bootstrap confidence intervals for performance metrics.
- bootReps
Number of bootstrap replications for confidence intervals.
- plotTree
Display visual representation of the optimal decision tree.
- plotComparison
Create comparison plots for different testing strategies.
- plotCostEffect
Create cost-effectiveness frontier plot.
- plotROC
Display ROC curves for top performing panels.
- prevalence
Known disease prevalence (0 = use sample prevalence).
- crossValidate
Perform k-fold cross-validation for panel performance.
- nFolds
Number of folds for cross-validation.
- seed
Random seed for reproducibility in bootstrap and cross-validation.
Value
A results object containing:
results$summary | a html | ||||
results$optimalPanel | a table | ||||
results$strategyComparison | a table | ||||
results$individualTests | a table | ||||
results$allCombinations | a table | ||||
results$treeStructure | a html | ||||
results$treePerformance | a table | ||||
results$crossValidation | a table | ||||
results$bootstrapResults | a table | ||||
results$treeVisualization | an image | ||||
results$strategyComparisonPlot | an image | ||||
results$costEffectivenessPlot | an image | ||||
results$rocCurvesPlot | an image | ||||
results$recommendations | a html |
Tables can be converted to data frames with asDF or as.data.frame. For example:
results$optimalPanel$asDF
as.data.frame(results$optimalPanel)