Compare Medical Decision Tests — decisioncompare • meddecide

Function for comparing multiple Medical Decision Tests. Compares sensitivity, specificity, positive predictive value, negative predictive value, and other metrics between different tests against the same golden standard. Includes statistical comparison using McNemar's test and confidence intervals for differences.

Usage

decisioncompare(
  data,
  gold,
  goldPositive,
  test1,
  test1Positive,
  test2,
  test2Positive,
  test3 = NULL,
  test3Positive,
  pp = FALSE,
  pprob = 0.3,
  od = FALSE,
  fnote = FALSE,
  ci = FALSE,
  plot = FALSE,
  radarplot = FALSE,
  statComp = FALSE,
  showSummary = FALSE,
  showExplanations = FALSE,
  showReportSentence = FALSE
)

Arguments

data: The data as a data frame.
gold: The gold standard reference variable representing true disease status.
goldPositive: The level indicating presence of disease in the gold standard variable.
test1: The first diagnostic test variable being evaluated for performance.
test1Positive: The level representing a positive result for Test 1.
test2: The second diagnostic test variable for comparison analysis.
test2Positive: The level representing a positive result for Test 2.
test3: Optional third diagnostic test variable for extended comparison analysis.
test3Positive: The level representing a positive result for Test 3.
pp: Enable custom prior probability (prevalence) for predictive value calculations.
pprob: Prior probability (disease prevalence in the community). Requires a value between 0.001 and 0.999, default 0.300.
od: Boolean selection whether to show frequency tables. Default is 'false'.
fnote: .
ci: .
plot: Generate comparison plot showing test performance metrics.
radarplot: Generate radar plot for comprehensive test comparison visualization.
statComp: Perform statistical comparison between tests (McNemar's test and confidence intervals for differences).
showSummary: Boolean to show natural language summary of statistical comparisons.
showExplanations: Boolean to show explanations, glossary, and educational content.
showReportSentence: Boolean to show manuscript-ready report sentence.

Value

A results object containing:

`results$text1`					a preformatted
`results$text2`					a html
`results$cTable1`					a table
`results$epirTable1`					a table
`results$cTable2`					a table
`results$epirTable2`					a table
`results$cTable3`					a table
`results$epirTable3`					a table
`results$comparisonTable`					a table
`results$mcnemarTable`					a table
`results$diffTable`					a table
`results$plot1`					an image
`results$plotRadar`					an image
`results$summaryReport`					a html
`results$reportSentence`					a html
`results$explanationsContent`					a html
`results$clinicalReport`					a html
`results$aboutAnalysis`					a html

Tables can be converted to data frames with asDF or as.data.frame. For example:

results$cTable1$asDF

as.data.frame(results$cTable1)

Examples

# Basic comparison of two diagnostic tests
library(ClinicoPath)
#> Error in library(ClinicoPath): there is no package called ‘ClinicoPath’
data('histopathology')

# Example 1: Compare imaging vs blood test performance
result1 <- decisioncompare(
    data = histopathology,
    gold = "Golden Standart",
    goldPositive = "1",
    test1 = "New Test",
    test1Positive = "1",
    test2 = "Rater 1",
    test2Positive = "1",
    ci = TRUE,
    plot = TRUE
)
#> Error in decisioncompare(data = histopathology, gold = "Golden Standart",     goldPositive = "1", test1 = "New Test", test1Positive = "1",     test2 = "Rater 1", test2Positive = "1", ci = TRUE, plot = TRUE): argument "test3Positive" is missing, with no default

# Example 2: Three-test comparison with statistical analysis
result2 <- decisioncompare(
    data = histopathology,
    gold = "Golden Standart",
    goldPositive = "1",
    test1 = "New Test",
    test1Positive = "1",
    test2 = "Rater 1",
    test2Positive = "1",
    test3 = "Rater 2",
    test3Positive = "1",
    statComp = TRUE,
    radarplot = TRUE
)
#> Warning: Removed 1 rows with missing values
#> Performing 3 pairwise comparisons with Holm-Bonferroni correction...
#> Statistical comparisons completed (3 comparisons, Holm-Bonferroni corrected).

# Example 3: Using custom prevalence for clinical setting
result3 <- decisioncompare(
    data = histopathology,
    gold = "Golden Standart",
    goldPositive = "1",
    test1 = "New Test",
    test1Positive = "1",
    test2 = "Rater 1",
    test2Positive = "1",
    pp = TRUE,
    pprob = 0.15,  # 15\% prevalence in screening population
    od = TRUE      # Show original frequency tables
)
#> Error in decisioncompare(data = histopathology, gold = "Golden Standart",     goldPositive = "1", test1 = "New Test", test1Positive = "1",     test2 = "Rater 1", test2Positive = "1", pp = TRUE, pprob = 0.15,     od = TRUE): argument "test3Positive" is missing, with no default