Package {ivdtools}

ivdtools: Statistical Tools for In Vitro Diagnostic Method Evaluation

Description

The package provides a collection of statistical workflows for evaluating in vitro diagnostic methods and reagents. Most multi-step workflows use S3 objects: construct an analysis object, add results with analysis generics, and inspect the accumulated results with summary() or plot().

Author(s)

Maintainer: hiox-tech GeorgeBinDragon@outlook.com [copyright holder]

See Also

Useful links:

• https://github.com/hiox-tech/ivdtools

• Report bugs at https://github.com/hiox-tech/ivdtools/issues

arrhenius — Arrhenius accelerated stability analysis

Description

Estimates degradation rates at multiple elevated temperatures, fits the Arrhenius equation, and extrapolates to the target storage temperature to predict shelf life.

Usage

arrhenius(
  data,
  temperature,
  time,
  value,
  target_temp,
  order = c("auto", "zero", "first"),
  direction = c("auto", "decrease", "increase"),
  limit,
  bias_type = c("relative", "absolute"),
  temp_unit = "C",
  time_unit = "d",
  conf.level = 0.95
)

Arguments

Value

An S3 object of class "arrhenius".

Examples

temp_df <- data.frame(temp = rep(c(25,30,37), each = 3),
                      time = rep(c(0,1,3), 3),
                      val = c(100,98,95, 100,96,90, 100,94,85))
arrhenius(temp_df, "temp", "time", "val", target_temp = 5, limit = 10)

Compute ROC curves and AUC for each evaluation column.

Description

Compute ROC curves and AUC for each evaluation column.

Usage

## S3 method for class 'roc'
auc(x, cols = NULL, ...)

Arguments

Value

Updated roc object with results stored in $auc_list and $auc_table

Examples

  obj <- roc(ivd_roc_example, cols = c("x1", "x2"), reference = "ref")
  obj <- auc(obj)
  obj <- auc(obj, cols = "x1")

Compare two paired ROC AUCs

Description

Compare two ROC curves evaluated on the common complete cases stored in a single roc() object. A curve can be a marker previously processed by auc() or a named multivariate model produced by mlr().

Usage

## S3 method for class 'roc'
auc_compare(
  x,
  curves,
  method = c("ep24", "delong"),
  conf.level = 0.95,
  rating_method = c("pearson", "kendall"),
  name = NULL,
  ...
)

Arguments

Details

method = "ep24" implements the paired Hanley–McNeil approximation in CLSI EP24-A2. It combines the two single-curve Hanley–McNeil standard errors with the correlation between the paired AUCs. The latter is obtained by linear interpolation of EP24-A2 Table 5 from the average within-group score correlation and average AUC. Coordinates outside the published table are bounded to its range and recorded in the result.

method = "delong" uses the nonparametric DeLong covariance estimator and handles ties with half credit. Both methods test the two-sided null hypothesis that the AUC difference is zero.

Value

The updated roc object, invisibly. The comparison is stored in x$auc_comparisons[[name]] as a roc_auc_comparison object.

Examples

obj <- roc(ivd_roc_example, c("x1", "x2"), "ref")
obj <- auc(obj)
obj <- auc_compare(obj, c("x1", "x2"), method = "ep24")
obj$auc_comparisons[[1L]]

obj <- mlr(obj, name = "combined")
obj <- auc_compare(obj, c("x1", "combined"), method = "delong")

Bias analysis (S3 method) – estimate bias at given medical decision levels

Description

Compute bias at specified medical decision levels (MDL) based on stored regression results: bias = predicted_candidate - reference = (intercept + slope * x0) - x0 = intercept + (slope - 1) * x0

Usage

## S3 method for class 'mcr'
bias(
  x,
  mdl,
  level = 0.95,
  interval = c("", "confidence", "prediction", "both"),
  ci_method = c("auto", "analytic", "jackknife", "rank", "bootstrap_percentile",
    "bootstrap_standard"),
  bootstrap_replicates = 5000L,
  seed = NULL,
  ...
)

Arguments

Details

Bias follows the conventional candidate-minus-reference direction; a positive value indicates that the candidate method is higher.

Value

Updated mcr object (invisible), results stored in $bias

Examples

  obj <- mcr(ivd_mcr_example, "sid", "test", "ref")
  obj <- regression(obj, method = "ols")
  bias(obj, mdl = c(200, 400, 600))
  bias(obj, mdl = c(200, 400, 600), interval = "both")

Bland-Altman analysis (S3 method)

Description

Calculate Limits of Agreement between two measurement methods, supporting three y-axis metrics: difference, ratio, and percent difference.

Usage

## S3 method for class 'mcr'
bland_altman(
  x,
  type = c("difference", "ratio", "percent"),
  x_axis = c("mean", "candidate", "reference"),
  percent_denominator = c("reference", "mean", "candidate"),
  agree.level = 0.95,
  conf.level = 0.95,
  method = c("parametric", "nonparametric", "hodges_lehmann"),
  median_ci_method = c("order_statistic", "interpolated"),
  bootstrap_replicates = 5000L,
  seed = NULL,
  ...
)

Arguments

Value

Updated mcr object (invisible), with results stored in $bland_altman. The result records the selected method, its centre estimate and center_ci. Method-specific aliases are mean_diff_ci for the parametric mean and median_diff_ci for the nonparametric median. The original mean_diff and sd_diff fields are retained for compatibility.

Examples

  obj <- mcr(ivd_mcr_example, "sid", "test", "ref")
  bland_altman(obj)
  bland_altman(obj, type = "ratio")
  bland_altman(obj, type = "percent", x_axis = "reference")
  bland_altman(obj, type = "percent", x_axis = "mean",
                percent_denominator = "mean")

bottle_anova constructor

Description

Create a bottle/batch ANOVA analysis object. Supports one-way and nested designs, with automatic assumption checks, missing value reporting, and ANOVA table output.

Usage

bottle_anova(data, formula, conf.level = 0.95)

Arguments

Value

An S3 object of class "bottle_anova" with components:

Examples

  # One-way design
  df1 <- data.frame(
    bottle = rep(c("A", "B", "C"), each = 5),
    value  = c(rnorm(5, 10, 1), rnorm(5, 12, 1), rnorm(5, 11, 1))
  )
  obj <- bottle_anova(df1, value ~ bottle)

  # Nested design
  df2 <- data.frame(
    lot   = rep(c("L1", "L2"), each = 10),
    vial  = rep(c("V1", "V2"), each = 5, times = 2),
    value = c(rnorm(5, 10, 1), rnorm(5, 10.5, 1),
              rnorm(5, 12, 1), rnorm(5, 12.5, 1))
  )
  obj2 <- bottle_anova(df2, value ~ lot/vial)

Estimate C5 and C95 from an internal continuous response

Description

Estimates response values associated with selected probabilities of a qualitative result. The response SD is modeled as a function of the expected response using a VFP precision profile. For direction = "geq", the event is result >= cutoff; for direction = "leq", the event is result <= cutoff. No acceptance decision is made.

Usage

c5_c95(
  data,
  result = NULL,
  sample = NULL,
  cutoff,
  direction = c("geq", "leq"),
  probabilities = c(0.05, 0.95),
  component = NULL,
  model.no = 1:10,
  K = 2,
  search_range = NULL,
  tol = 1e-07
)

Arguments

Value

An object of class c5_c95. Its estimates element contains the estimated response, fitted SD, achieved probability, extrapolation flag, and root-search diagnostics for each cutoff and probability.

Examples

set.seed(1203)
cx_data <- data.frame(
  level = rep(paste0("L", 1:7), each = 20),
  response = unlist(lapply(seq(34, 46, length.out = 7), function(mu)
    stats::rnorm(20, mean = mu, sd = 0.7 + 0.01 * mu)))
)
cx <- c5_c95(cx_data, result = "response", sample = "level",
             cutoff = 40, model.no = 1)
cx$estimates
plot(cx)

S3 confidence interval method

Description

Compute confidence intervals for each variance component (SD and %CV) based on Satterthwaite approximation. Requires variance() to be run first to populate the $results slot.

Usage

## S3 method for class 'precision'
ci(x, digits = 4, ...)

Arguments

Details

This method reports intervals for the fitted model components and the total component. EP05 semantic combinations such as multisite within-laboratory precision require the day, run, and site mapping supplied to report.precision and are therefore constructed by report(x, ..., table = "CI").

Value

Updated precision object with results stored in $ci

Examples

  data(VCAdata1, package = "VCA")
  obj <- precision(VCAdata1, y ~ day/run, by = "sample")
  obj <- variance(obj)
  obj <- ci(obj)

Extract coefficients from an equation fit

Description

Extract coefficients from an equation fit

Usage

## S3 method for class 'fit_equation'
coef(object, ...)

Arguments

Value

A named numeric vector of model coefficients.

Examples

df <- data.frame(x = 1:5, y = c(2.1, 4.0, 6.2, 7.9, 10.1))
f <- fit_equation("E01", df, "x", "y")
coef(f)

Assess reference-material commutability using four standard approaches

Description

Implements the statistical calculations for CLSI EP14-A3 Deming-regression prediction intervals, CLSI EP30-A prediction intervals and relative residuals, the IFCC Part 2 difference-in-bias approach, or the IFCC Part 3 calibration-effectiveness approach. The function expects long-format data and does not assess specimen collection, run randomization, reagent lots, or other experimental conduct.

Usage

commutability(
  data,
  sample_id,
  material_type = NULL,
  procedure,
  result,
  position = NULL,
  approach = c("ep14", "ep30", "ifcc", "calibration"),
  scale = c("raw", "log10", "ln"),
  reference_procedure = NULL,
  procedure_pairs = NULL,
  conf.level = 0.95,
  criterion = NULL,
  bias_model = c("constant_global", "constant_local", "linear_local"),
  local_n = NULL,
  coverage_factor = 1.9,
  rm_position = c("pooled", "material_specific"),
  rm_set_size = 1L,
  relative_residual_limit = 2,
  calibration_stage = NULL,
  calibration_replicate_summary = c("mean", "median", "error"),
  exclude_procedures = NULL,
  w3_reference_procedure = NULL,
  material_type_map = NULL
)

Arguments

Details

Sample-size, replicate-count, position-matching, and method-count targets from CLSI and IFCC publications are reported in design_checks; they are not treated as proof of compliance and do not stop a calculation that is statistically defined. Conditions required by the formula itself still produce an error. A local result that cannot be calculated is returned as not_evaluable, while an unavailable secondary statistic is returned as NA.

Value

An object of class commutability_result containing material-level results, model estimates, design checks, IFCC position diagnostics, and the transformed analysis data. Design checks report whether the supplied study meets the sample-size and replication recommendations of the named method; unmet checks do not prevent otherwise computable analyses.

Examples

# Long-format replicate results for eight clinical samples.
clinical <- expand.grid(
  sample = paste0("CS", 1:8), procedure = c("MP1", "MP2"),
  replicate = 1:3, stringsAsFactors = FALSE
)
level <- setNames(seq(10, 80, length.out = 8), paste0("CS", 1:8))
clinical$result <- level[clinical$sample] +
  ifelse(clinical$procedure == "MP2", 1, 0) +
  rep(c(-0.2, 0, 0.2), each = 16)
clinical$material_type <- "clinical"
clinical$position <- NA_integer_

# One reference material measured at three positions, in triplicate.
rm <- expand.grid(
  sample = "RM1", procedure = c("MP1", "MP2"),
  position = 1:3, replicate = 1:3, stringsAsFactors = FALSE
)
rm$result <- 45 + ifelse(rm$procedure == "MP2", 1, 0) +
  rep(c(-0.1, 0, 0.1), each = 6)
rm$material_type <- "rm"
columns <- c("sample", "material_type", "procedure", "result",
             "position", "replicate")
example_data <- rbind(clinical[columns], rm[columns])

ep14 <- commutability(
  example_data, "sample", "material_type", "procedure", "result",
  approach = "ep14"
)
ep14$results

# Explicitly map nonstandard material-type labels.
mapped_data <- example_data
mapped_data$material_type <- ifelse(
  mapped_data$material_type == "clinical", "patient_sample", "candidate_rm"
)
ep14_mapped <- commutability(
  mapped_data, "sample", "material_type", "procedure", "result",
  approach = "ep14",
  material_type_map = c(patient_sample = "clinical", candidate_rm = "rm")
)
ep14_mapped$results

ifcc <- commutability(
  example_data, "sample", "material_type", "procedure", "result",
  position = "position", approach = "ifcc", criterion = 2,
  bias_model = "constant_global"
)
ifcc$results

Fit and compare multiple equations

Description

Fit and compare multiple equations

Usage

compare_equation(data, x = "x", y = "y", eqs = NULL, ...)

Arguments

Value

An object of class "compare_equation" with components:

Examples

df <- data.frame(
  x = c(0.1, 0.3, 1, 3, 10, 30, 100),
  y = c(12, 25, 48, 72, 88, 96, 99)
)
compare_equation(df, "x", "y", eqs = c("E01", "E06", "E07"))

Convert quantitative columns to binary based on cutoff values

Description

Creates new binary columns (0/1) from existing quantitative columns using specified cutoff values, appending them to the original data frame. New columns are named <col>_binary.

Usage

continuous_to_binary(data, cols, cutoff)

Arguments

Value

data frame with original columns plus new <col>_binary columns

Examples

  df <- data.frame(x = 1:10, y = rnorm(10))
  continuous_to_binary(df, "x", 5)
  continuous_to_binary(df, c("x", "y"), c(5, 0))

Correlation analysis (S3 method)

Description

Perform correlation analysis between the reference method (X) and candidate method (Y), supporting Pearson, Spearman, and Kendall methods.

Usage

## S3 method for class 'mcr'
correlation(x, method = c("pearson", "spearman", "kendall"), ...)

Arguments

Value

Updated mcr object (invisible), results stored in $correlation

Examples

  obj <- mcr(ivd_mcr_example, "sid", "test", "ref")
  correlation(obj)
  correlation(obj, method = "spearman")

Construct a 2x2 contingency table from TP / FP / TN / FN frequencies (silent construction)

Description

When raw data are not available and only the four diagnostic test frequencies are known, use this function to directly construct a result object with the same structure as raw_to_table(), compatible with print() for displaying the fourfold table.

Usage

counts_to_table(
  tp,
  fp,
  tn,
  fn,
  candidate = "candidate",
  reference = "reference",
  candidate_levels = c("positive", "negative"),
  reference_levels = c("positive", "negative")
)

Arguments

Value

A fourfold_table object (S3 class "fourfold_table") with the same structure as raw_to_table(): - $freq_raw Long-format frequency data.frame - $table 3x3 matrix with margins - $n Total sample size - $print Print slot - $candidate_levels Levels of the candidate method - $reference_levels Levels of the reference method Use print() directly to display the fourfold table.

Examples

  result <- counts_to_table(tp = 45, fp = 12, tn = 80, fn = 5,
                          candidate = "new method", reference = "gold standard")

Optimal Cutoff Analysis (S3 method)

Description

For each evaluation column, compute diagnostic metrics at all cutoff values, and identify the optimal cutoff (maximum Youden index, closest to (0,1) corner).

Usage

## S3 method for class 'roc'
cutoff(x, cols = NULL, ...)

Arguments

Value

Updated roc object with results stored in $cutoff_table

Examples

  obj <- roc(ivd_roc_example, cols = c("x1", "x2"), reference = "ref")
  obj <- auc(obj)
  cutoff(obj)
  cutoff(obj, cols = "x1")

Describe an analysis object

Description

Describe an analysis object

Usage

describe(x, ...)

correlation(x, ...)

regression(x, ...)

bias(x, ...)

bland_altman(x, ...)

report(x, ...)

profile(object, ...)

normal(x, ...)

ci(x, ...)

variance(x, ...)

vc(x, ...)

diagnostics(object, ...)

kappa(x, ...)

mcnemar(x, ...)

cutoff(x, ...)

mlr(x, ...)

auc(x, ...)

auc_compare(x, ...)

tukey(x, ...)

Arguments

Value

The value returned by the dispatched method.

The value returned by the dispatched method.

The value returned by the dispatched method.

The value returned by the dispatched method.

The value returned by the dispatched method.

The value returned by the dispatched method.

The value returned by the dispatched method.

The value returned by the dispatched method.

The value returned by the dispatched method.

The value returned by the dispatched method.

The value returned by the dispatched method.

The value returned by the dispatched method.

The value returned by the dispatched method.

The value returned by the dispatched method.

The value returned by the dispatched method.

The value returned by the dispatched method.

The value returned by the dispatched method.

The value returned by the dispatched method.

The value returned by the dispatched method.

Examples

obj <- mcr(ivd_mcr_example, "sid", "test", "ref")
describe(obj)

Descriptive Statistics: Raw Data Overview (S3 Method)

Description

Print variable description information for a fourfold_table object (from raw_to_table). If the object comes from counts_to_table (no raw data), indicates unavailability.

Usage

## S3 method for class 'fourfold_table'
describe(x, ...)

Arguments

Value

Updated fourfold_table object, with $describe slot filled with description results

Examples

  df <- data.frame(
    new    = c("positive","positive","negative","negative","positive","negative"),
    gold   = c("positive","negative","positive","negative","positive","positive"),
    age    = c(45, 52, 38, 61, 47, 55),
    gender = c("M","F","F","M","M","F"),
    id     = c("S001","S002","S003","S004","S005","S006")
  )
  tab <- raw_to_table(df, "new", "gold", id = "id")
  tab <- describe(tab)

Compute descriptive statistics and store results

Description

Computes data overview (rows/columns/ID duplicates), per-column summaries, and a side-by-side comparison table of candidate, reference, and Diff. Results are stored in x$describe (class mcr_describe) and displayed via print.mcr_describe() (e.g., when calling summary()).

Usage

## S3 method for class 'mcr'
describe(x, cols = NULL, digits = 4L, ...)

Arguments

Value

Updated mcr object (invisible), results stored in x$describe

Examples

  obj <- mcr(ivd_mcr_example, "sid", "test", "ref")
  describe(obj)

Descriptive Statistics (S3 method)

Description

Print distribution summaries for evaluation columns and the reference column (if numeric). When the reference is binary, evaluation column distributions are shown grouped by positive/negative.

Usage

## S3 method for class 'roc'
describe(x, cols = NULL, digits = 4L, ...)

Arguments

Value

Updated roc object with results stored in $describe

Examples

  obj <- roc(ivd_roc_example, cols = "x1", reference = "ref")
  describe(obj)

Diagnostic performance metrics: calculate sensitivity, specificity, likelihood ratios, etc. with confidence intervals for a fourfold_table

Description

Diagnostic performance metrics: calculate sensitivity, specificity, likelihood ratios, etc. with confidence intervals for a fourfold_table

Usage

## S3 method for class 'fourfold_table'
diagnostics(
  object,
  conf.level = 0.95,
  ci.method = "wilson",
  prevalence = NULL,
  ...
)

Arguments

Value

Updated fourfold_table object, with diagnostic performance metrics stored in $diagnostics: - $tp / $fp / $fn / $tn / $n Four-fold frequencies and total sample size - $sensitivity Sensitivity with CI - $specificity Specificity with CI - $ppv Positive predictive value with CI - $npv Negative predictive value with CI - $accuracy Accuracy with CI - $prevalence Prevalence with CI - $lr_positive Positive likelihood ratio with CI - $lr_negative Negative likelihood ratio with CI - $odds_ratio Odds ratio with CI - $conf_level Confidence level used - $ci_method Method name used - $prev_specified Whether user specified prevalence

Examples

  tab <- raw_to_table(ivd_qualitative_example, "new", "gold",
                      positive = "positive", id = "id")
  tab <- diagnostics(tab)
  tab <- diagnostics(tab, prevalence = 0.3)
  tab <- diagnostics(tab, conf.level = 0.90)
  tab <- diagnostics(tab, ci.method = "clopper-pearson")

Calculate dilution recovery

Description

Summarizes replicate measurements and calculates target concentration, dilution-corrected concentration, and recovery for each specimen and dilution factor. Across-specimen mean recovery and its Student t confidence interval are calculated for every dilution factor. No acceptance criterion or suitability decision is applied.

Usage

dilution_recovery(
  data,
  sample,
  result,
  dilution_factor,
  assigned = NULL,
  diluent_value = 0,
  conf.level = 0.95
)

Arguments

Value

An object of class dilution_recovery containing results, summary, the retained data, and excluded row indices.

References

CLSI. Establishing and Verifying an Extended Measuring Interval Through Specimen Dilution and Spiking. EP34, 1st ed. 2018.

Examples

dilution_data <- data.frame(
  sample = rep(c("S1", "S2"), each = 6),
  factor = rep(rep(c(1, 2, 4), each = 2), 2),
  result = c(99, 101, 49, 51, 24, 26,
             199, 201, 99, 101, 49, 51)
)
dilution <- dilution_recovery(
  dilution_data, "sample", "result", "factor"
)
dilution$summary

General-purpose equation fitting

Description

General-purpose equation fitting

Usage

fit_equation(
  eq,
  data,
  x = "x",
  y = "y",
  start = NULL,
  lower = NULL,
  upper = NULL,
  constraints = NULL,
  weights = NULL,
  ...
)

Arguments

Value

an S3 object of class "fit_equation"

Examples

# ==== Response curve fitting ====

# 1) Linear
df <- data.frame(x = 1:5, y = c(2.1, 4.0, 6.2, 7.9, 10.1))
f1 <- fit_equation("E01", df, "x", "y")
print(f1)

# 2) Logarithmic
f2 <- fit_equation("E06", df, "x", "y")
print(f2)

# 3) 4PLC (dose-response, descending)
df4plc <- data.frame(
  conc = c(0.1, 0.3, 1, 3, 10, 30, 100),
  resp = c(99, 96, 88, 72, 48, 25, 12)
)
f3 <- fit_equation("E07", df4plc, "conc", "resp")
print(f3)

# 4) Custom formula
f4 <- fit_equation("a + b * log(x)", df4plc, "conc", "resp")
print(f4)

# 5) Box constraints (parameter bounds)
f5 <- fit_equation("E07", df4plc, "conc", "resp",
  lower = list(D = 0), upper = list(D = 5))
print(f5)

# 6) Inequality constraints (parameter relationship)

# require D > 1 (Hill slope cannot be too shallow)
f6 <- fit_equation("E07", df4plc, "conc", "resp",
  constraints = list(ineq = function(p) 1 - p["D"]))
print(f6)


# 7) Equality constraints (fixed parameter relationship)

# fix b = 0.5 in exponential decay y = a*exp(-b*x)
df_exp <- data.frame(x = 0:5, y = c(10, 6.1, 3.7, 2.2, 1.4, 0.8))
f7 <- fit_equation("E04", df_exp, "x", "y",
  constraints = list(eq = function(p) p["b"] - 0.5))
print(f7)


# 8) Weighted fitting via replicate_to_mean

# data with replicate measurements (8 dose levels, 3 replicates each)
df_rep <- data.frame(
  dose = rep(c(0.3, 0.6, 1.5, 4, 10, 25, 60, 150), each = 3),
  resp = c(4.1, 3.8, 3.9, 8.5, 8.9, 8.6,
           18.2, 17.9, 18.5,
           34.6, 35.1, 34.8,
           54.3, 53.9, 54.7,
           73.0, 73.6, 73.2,
           87.5, 87.1, 87.8,
           97.9, 98.3, 98.0)
)
# first summarize, generate inverse-variance weights
rep <- replicate_to_mean(df_rep, "dose", "resp", weights = "1/sd^2")

# fit 4PLC with summarized means and explicit start values
f8 <- fit_equation("E07", rep, x = "x", y = "y_mean",
  weights = rep$weight,
  start = list(A = 2, B = 100, C = 8, D = -1.5))
print(f8)

Summarize a high-dose hook-effect experiment

Description

Summarizes raw response by known concentration and identifies the first post-peak measured concentration whose mean response is at or below the raw response associated with the ULoQ. The reported hook_concentration is the immediately preceding measured concentration; no interpolation is used. No hook-effect or extended-interval acceptability decision is made.

Usage

hook_effect(
  data,
  concentration,
  response,
  series = NULL,
  uloq,
  uloq_response = NULL,
  conf.level = 0.95
)

Arguments

Value

An object of class hook_effect containing concentration-level summaries and one threshold result per series.

References

CLSI. Establishing and Verifying an Extended Measuring Interval Through Specimen Dilution and Spiking. EP34, 1st ed. 2018.

Examples

hook_data <- data.frame(
  concentration = rep(c(100, 200, 400, 800), each = 2),
  response = c(48, 52, 98, 102, 148, 152, 88, 92)
)
hook <- hook_effect(
  hook_data, "concentration", "response", uloq = 200
)
hook$results

Analyze an EP07 interference dose-response experiment

Description

Summarizes replicate results at each interferent concentration relative to an observed baseline concentration. The recommended point-to-point method is the default; a straight-line regression can be requested explicitly.

Usage

interference_dose_response(
  data,
  result,
  concentration,
  by = NULL,
  baseline = 0,
  method = c("point_to_point", "linear"),
  conf.level = 0.95
)

Arguments

Value

An object of class interference_dose_response.

Examples

dose_data <- expand.grid(
  interferent = c("A", "B"), concentration = c(0, 10, 20, 30, 40),
  replicate = 1:5, KEEP.OUT.ATTRS = FALSE
)
dose_data$result <- 100 + ifelse(dose_data$interferent == "A", 0.5, -0.3) *
  dose_data$concentration + stats::rnorm(nrow(dose_data), 0, 0.5)
dose_fit <- interference_dose_response(
  dose_data, "result", "concentration", by = "interferent")
dose_fit$summary

Analyze paired-difference interference data

Description

Supports the standard EP07 test/control preparation and two laboratory extensions: analyte spiking in normal/interferent matrices and direct measurement of interferent dissolved in a blank. One row represents one replicate result. Multiple interferents or measurand levels are handled by supplying their columns in by.

Usage

interference_paired(
  data,
  result,
  design = c("ep07", "analyte_spike", "blank"),
  by = NULL,
  condition = NULL,
  test_level = "test",
  control_level = "control",
  matrix = NULL,
  normal_level = "normal",
  interferent_level = "interferent",
  spike = NULL,
  unspiked_level = "unspiked",
  spiked_level = "spiked",
  conf.level = 0.95
)

Arguments

Value

An S3 object of class interference_paired containing effects, cell_summary, issues, and excluded row indices.

Examples

ep07_data <- data.frame(
  interferent = rep(c("bilirubin", "hemoglobin"), each = 10),
  condition = rep(rep(c("control", "test"), each = 5), 2),
  result = c(0.45, 0.56, 0.48, 0.54, 0.47,
             0.24, 0.28, 0.35, 0.37, 0.26,
             5.02, 4.98, 5.08, 4.95, 5.00,
             5.20, 5.12, 5.18, 5.10, 5.15)
)
interference_paired(ep07_data, "result", design = "ep07",
                    condition = "condition", by = "interferent")

spike_data <- expand.grid(
  matrix = c("normal", "interferent"), spike = c("before", "after"),
  replicate = 1:4, KEEP.OUT.ATTRS = FALSE
)
spike_data$result <- with(spike_data,
  ifelse(matrix == "normal", 10, 12) +
  ifelse(spike == "after", ifelse(matrix == "normal", 10, 8), 0))
interference_paired(spike_data, "result", design = "analyte_spike",
                    matrix = "matrix", spike = "spike",
                    unspiked_level = "before", spiked_level = "after")

Evaluate interference with patient specimens

Description

Averages replicates within specimen and method, then calculates the evaluated procedure result minus the comparative procedure result. Potential interferent concentrations can be supplied as multiple numeric columns and are modeled separately.

Usage

interference_patient(
  data,
  id,
  group,
  method,
  result,
  test_group,
  control_group,
  evaluated_method,
  comparative_method,
  interferent_cols = NULL,
  conf.level = 0.95
)

Arguments

Value

An object of class interference_patient.

Examples

patient_data <- expand.grid(
  id = paste0("S", 1:12), method = c("evaluated", "comparative"),
  replicate = 1:2, KEEP.OUT.ATTRS = FALSE
)
patient_data$group <- ifelse(as.integer(sub("S", "", patient_data$id)) <= 6,
                             "control", "test")
patient_data$bilirubin <- 2 * (as.integer(sub("S", "", patient_data$id)) - 1)
truth <- 50 + as.integer(sub("S", "", patient_data$id))
patient_data$result <- truth +
  ifelse(patient_data$method == "evaluated", 0.08 * patient_data$bilirubin, 0)
patient_fit <- interference_patient(
  patient_data, "id", "group", "method", "result",
  test_group = "test", control_group = "control",
  evaluated_method = "evaluated", comparative_method = "comparative",
  interferent_cols = "bilirubin")
patient_fit$group_summary

Calculate replicate counts for an EP07 paired-difference experiment

Description

Calculates the number of replicate measurements needed in each of the test and control samples. sd and detectable_difference must use the same scale: both absolute units, or both relative units (SD/CV and percent). The calculation is vectorized and reports design parameters only.

Usage

interference_replicates(
  sd,
  detectable_difference,
  alpha = 0.05,
  power = 0.9,
  alternative = c("two.sided", "one.sided"),
  min_replicates = 5L,
  label = NULL
)

Arguments

Value

A data frame of class interference_replicates.

References

CLSI. Interference Testing in Clinical Chemistry. EP07, 3rd ed.

Examples

interference_replicates(sd = c(5, 5),
                        detectable_difference = c(7, 10),
                        label = c("7 percent", "10 percent"))

Small deterministic one-way ANOVA example

Description

Small deterministic one-way ANOVA example

Usage

ivd_bottle_example

Format

A data frame with 12 observations from three bottles.

Small deterministic method-comparison example

Description

Small deterministic method-comparison example

Usage

ivd_mcr_example

Format

A data frame with 12 rows and columns sid, test, and ref.

Small deterministic qualitative-method example

Description

Small deterministic qualitative-method example

Usage

ivd_qualitative_example

Format

A data frame with paired binary results and sample identifiers.

Small deterministic ROC example

Description

Small deterministic ROC example

Usage

ivd_roc_example

Format

A data frame with 12 rows, two markers, and a binary reference.

Cohen's Kappa / PABAK Agreement Analysis (2x2 table only)

Description

Calculate Cohen's Kappa coefficient or PABAK (Prevalence-Adjusted Bias-Adjusted Kappa) and its confidence interval, used to evaluate agreement between two binary classification methods.

Usage

## S3 method for class 'fourfold_table'
kappa(x, prevalence = NULL, conf.level = 0.95, ...)

Arguments

Details

When data prevalence is extreme, Cohen's Kappa may be low even when observed agreement is high. In such cases, specify the prevalence parameter to compute PABAK as a correction. PABAK = 2 x p_obs - 1, which assumes expected agreement = 0.5.

Value

Updated fourfold_table object, with Kappa results stored in $kappa: - $kappa Kappa / PABAK coefficient - $se Standard error - $lower / $upper Lower/upper confidence interval bounds - $p_obs Observed agreement - $p_exp Expected agreement - $n Total sample size - $conf_level Confidence level used - $method Method name used

Examples

  tab <- counts_to_table(tp = 85, fp = 3, tn = 90, fn = 2,
                         candidate = "new method", reference = "gold standard")
  tab <- kappa(tab)                        # Cohen's Kappa
  tab <- kappa(tab, prevalence = 0.5)      # PABAK

Evaluate linearity using straight-line or polynomial regression

Description

linearity() is a parameter-driven analysis engine rather than a separate validation/verification workflow. The straight-line calculations implement the approach in CLSI EP06-Ed2. Polynomial regression reproduces the superseded EP6-A procedure and is reported as exploratory. Straight-line models are fitted to sample-level means; EP6-A polynomial models are fitted to the retained replicate observations.

Usage

linearity(
  data,
  sample,
  result,
  x,
  method = c("linear", "polynomial"),
  intercept = TRUE,
  weights = c("equal", "level_variance", "pooled", "precision_profile"),
  variance_group = NULL,
  max_degree = 3L,
  poly_alpha = 0.05,
  percent_base = c("predicted", "x", "mean"),
  conf.level = NULL,
  multiplicity = c("familywise", "none"),
  adl = NULL,
  adl_abs = NULL,
  adl_rel = NULL,
  profile_exclude_low = TRUE,
  profile_intercept = FALSE
)

Arguments

Details

With weights = "pooled", variance groups must be supplied explicitly. Within group g, the pooled variance is sum((n[j] - 1) * s[j]^2) / sum(n[j] - 1), and the level-mean regression weight is n[j] / pooled_variance[g]. The function does not infer groups from the number of concentration levels because EP06 grouping also depends on the observed repeatability pattern.

Value

An S3 object of class linearity.

Examples

set.seed(2020)
linearity_data <- data.frame(
  level = rep(1:6, each = 4),
  expected = rep(c(10, 30, 50, 70, 90, 110), each = 4)
)
linearity_data$result <- 2 + 0.98 * linearity_data$expected +
  stats::rnorm(nrow(linearity_data), sd = 1.5)

# Parameter-driven straight-line analysis.
fit_linear <- linearity(
  linearity_data, sample = "level", result = "result", x = "expected",
  method = "linear", intercept = TRUE, weights = "level_variance",
  adl_rel = 5
)
fit_linear$model_comparison
fit_linear$level_summary

# EP6-A polynomial analysis (reported as exploratory under EP06-Ed2).
fit_polynomial <- linearity(
  linearity_data, sample = "level", result = "result", x = "expected",
  method = "polynomial", max_degree = 3
)
fit_polynomial$coefficients

Calculate target LOW and HIGH concentrations for verification

Description

Implements the equations in Appendix J of CLSI EP06-Ed2 for studies in which results outside the measuring interval cannot be obtained.

Usage

linearity_endpoints(
  lloq = NULL,
  uloq = NULL,
  cv_high = NULL,
  cv_at_lloq = NULL,
  cv_at_k_lloq = NULL,
  k = 3,
  cv_scale = c("percent", "proportion")
)

## S3 method for class 'linearity_endpoints'
print(x, ...)

Arguments

Value

A list of class linearity_endpoints.

The unchanged x, invisibly.

Examples

linearity_endpoints(
  lloq = 10, uloq = 1000,
  cv_high = 5, cv_at_lloq = 20, cv_at_k_lloq = 15
)

Design a HIGH/LOW linearity panel

Description

Design a HIGH/LOW linearity panel

Usage

linearity_panel(
  proportion_high,
  high,
  low = 0,
  total_volume = NULL,
  sample = NULL
)

## S3 method for class 'linearity_panel'
print(x, ...)

Arguments

Value

A data frame containing proportions, expected values, relative concentrations, and (when requested) component volumes.

The unchanged x, invisibly.

Examples

linearity_panel(
  proportion_high = c(1, 0.75, 0.5, 0.25, 0),
  high = 210, low = 10, total_volume = 1
)

Calculate replicate counts for a linearity study

Description

Calculates the minimum number of replicates from the imprecision and allowable deviation from linearity (ADL), following CLSI EP06-Ed2. Relative inputs use percentage units (for example, 5 means 5%). Absolute inputs must use the same measurement unit for imprecision, adl, and true_deviation.

Usage

linearity_replicates(
  imprecision,
  adl,
  type = c("relative", "absolute"),
  level_probability = 0.99,
  levels = NULL,
  family_error = NULL,
  true_deviation = 0,
  min_replicates = 2L
)

## S3 method for class 'linearity_replicates'
print(x, ...)

Arguments

Value

A data frame of class linearity_replicates.

The unchanged x, invisibly.

Examples

# Relative ADL: both CV and ADL are expressed in percent.
linearity_replicates(imprecision = c(3, 5), adl = 5)

# Choose the per-level probability from a 10% family-wise failure risk.
linearity_replicates(5, 5, levels = 9, family_error = 0.10)

Print the equation registry

Description

Print the equation registry

Usage

list_equation(category = NULL, engine = NULL)

Arguments

Value

invisible data.frame (matching rows), prints a formatted table to the console

Examples

# All equations
list_equation()

# Response only
list_equation(category = "Response")

# lm engine only
list_equation(engine = "lm")

List Sadler precision profile model equations

Description

List Sadler precision profile model equations

Usage

list_sadler()

Arguments

This function has no arguments.

Details

List the 10 candidate Sadler precision profile model formulas and types used in the VFP package. Includes: Constant SD, Constant CV, Linear (variance), Power model, Exponential model, etc.

Value

Invisibly returns NULL. The model table is printed to the console as a side effect.

Examples

  list_sadler()

List Westgard QC rules

Description

Prints commonly used Westgard multi-rule QC rules and their meanings. Rule names printed here can be passed to qc_chart(rules = "...").

Usage

list_westgard(brief = FALSE)

Arguments

Value

A named character vector of rule descriptions (invisibly).

Examples

  list_westgard()

McNemar Test (for fourfold_table only)

Description

Perform McNemar's test on a 2x2 paired contingency table to evaluate whether there is a systematic difference between two binary classification methods (i.e., whether marginal probabilities are equal).

Usage

## S3 method for class 'fourfold_table'
mcnemar(x, conf.level = 0.95, ...)

Arguments

Details

When discordant pairs (b + c) are few (< 25), automatically uses the exact binomial test (binom.test); otherwise uses McNemar's chi-squared test with continuity correction.

Value

Updated fourfold_table object, with McNemar results stored in $mcnemar: - $chi_sq McNemar chi-squared statistic (NA for exact test) - $df Degrees of freedom (NA for exact test) - $p_value Test p-value - $method Test method used - $b Discordant pairs (positive, negative) count - $c Discordant pairs (negative, positive) count - $n_bc Total discordant pairs - $conf.level Confidence level used

Examples

  tab <- counts_to_table(tp = 45, fp = 12, tn = 80, fn = 5,
                         candidate = "new method", reference = "gold standard")
  tab <- mcnemar(tab)

mcr constructor

Description

Create a method comparison regression (mcr) object containing one summarized row per sample ID. Raw replicate rows must first be summarized with replicate_to_mean() or by the user.

Usage

mcr(
  data,
  id,
  candidate,
  reference,
  weights = NULL,
  candidate_sd = NULL,
  reference_sd = NULL,
  candidate_n = NULL,
  reference_n = NULL
)

Arguments

Value

Returns an S3 "mcr" object containing:

and analysis result storage slots: $correlation, $regression, $outlier, $bland_altman

Examples

  df <- data.frame(sid = 1:30,
                   test = rnorm(30, 50, 10),
                   ref  = rnorm(30, 50, 10))
  obj <- mcr(df, "sid", "test", "ref")

mkt — Mean Kinetic Temperature

Description

Calculates the mean kinetic temperature from one or more temperature probes. Supports equal-interval and time-weighted trapezoidal methods.

Usage

mkt(data, temp_cols, time = NULL, temp_unit = "C", ea = 83.144)

Arguments

Value

An S3 object of class "mkt".

Examples

temp_df <- data.frame(t1 = c(25,26,27), t2 = c(24,25,26), time = 1:3)
mkt(temp_df, c("t1","t2"), "time")

Multivariate Logistic Regression (S3 method)

Description

Fit logistic regression using selected evaluation columns, compute AUC of predicted probabilities. Names must be unique; auto-generated as MLR01, MLR02 ... when not specified.

Usage

## S3 method for class 'roc'
mlr(x, cols = NULL, name = NULL, ...)

Arguments

Value

Updated roc object with results stored in $mlr_results[name]

Examples

  obj <- roc(ivd_roc_example, cols = c("x1", "x2"), reference = "ref")
  obj <- auc(obj)
  obj <- mlr(obj)                     # -> MLR01
  obj <- mlr(obj)                     # -> MLR02
  obj <- mlr(obj, name = "my_model")  # -> my_model

S3 normality test method

Description

Perform normality tests on the response values for each sample, underlying call to normal_test(). Supports automatic test selection (Shapiro-Wilk, Anderson-Darling, Lilliefors, CVM).

Usage

## S3 method for class 'precision'
normal(
  x,
  method = c("auto", "shapiro", "sw", "ad", "lillie", "cvm"),
  level = 0.95,
  ...
)

Arguments

Value

Updated precision object with results stored in $normal. Each sample entry contains test_method, statistic_name, statistic, p_value, and n; W is retained as a legacy alias only for Shapiro-Wilk results.

Examples

  data(VCAdata1, package = "VCA")
  obj <- precision(VCAdata1, y ~ day/run, by = "sample")
  obj <- normal(obj)

Normality test

Description

Performs normality tests on a single numeric column of a data frame. Plots are generated separately via plot().

Usage

normal_test(
  data,
  col,
  method = c("auto", "shapiro", "sw", "ad", "lillie", "cvm"),
  level = 0.95
)

Arguments

Value

An S3 object of class "normal_test" containing:

call

Function call.

data_name

Data object name.

col

Column name.

method

User-specified method.

level

Confidence level.

n_total

Total number of rows.

n

Number of finite values (non-missing).

missing

Number of missing values.

test_method

Actual test method used.

statistic

Test statistic.

p_value

p-value.

y

Clean numeric vector, used by plot().

Examples

  df <- data.frame(x = rnorm(50))
  normal_test(df, "x")
  normal_test(df, "x", method = "ad")
  plot(normal_test(df, "x"))
  plot(normal_test(df, "x"), type = "his")
  plot(normal_test(df, "x"), type = c("qq", "his"))

Detect outliers in an analysis object

Description

Detect outliers in an analysis object

Usage

outlier(x, ...)

Arguments

Value

The value returned by the dispatched method.

Examples

  obj <- mcr(ivd_mcr_example, "sid", "test", "ref")
  outlier(obj, method = "grubbs")

Outlier detection (S3 method)

Description

Perform outlier detection based on the difference or percent difference between two measurements. Reuses four methods from ../precision/outliers.R.

Usage

## S3 method for class 'mcr'
outlier(
  x,
  method = c("grubbs", "esd", "dixon", "iqr"),
  type = c("difference", "percent"),
  percent_denominator = c("reference", "mean", "candidate"),
  alpha = 0.05,
  coef = 1.5,
  ...
)

Arguments

Value

Updated mcr object (invisible), results stored in $outlier

Examples

  obj <- mcr(ivd_mcr_example, "sid", "test", "ref")
  outlier(obj, method = "grubbs")
  outlier(obj, method = "iqr", type = "percent")

S3 outlier detection method

Description

Detect outliers in the response values for each sample, supporting Grubbs test and IQR method. Underlying call to outliers_test() (from outliers-and-normal.R).

Usage

## S3 method for class 'precision'
outlier(x, alpha = 0.05, method = c("grubbs", "iqr"), ...)

Arguments

Value

Updated precision object with results stored in $outlier

Examples

  data(VCAdata1, package = "VCA")
  obj <- precision(VCAdata1, y ~ day/run, by = "sample")
  obj <- outlier(obj, method = "iqr")

Outlier detection

Description

Integrates Grubbs test, generalized ESD test, Dixon Q test, and IQR method under a consistent API with an S3 print method.

Usage

outliers_test(data, col, method = c("grubbs", "esd", "dixon", "iqr"), ...)

Arguments

Value

An S3 object of class "outliers_test" containing:

method

Method name used.

data_name

Name of the input data.

col

Column name tested.

n_total

Total number of rows.

n

Number of finite observations used.

missing

Number of missing (non-finite) values.

parameters

List of method parameters.

n_outliers

Number of outliers detected.

indices

Indices of outliers in the original vector (1-based).

values

Outlier values.

details

Method-specific details (statistics, critical values, bounds, etc.).

Examples

set.seed(123)
df <- data.frame(x = c(rnorm(20), 10, -8))
outliers_test(df, "x", "grubbs")
outliers_test(df, "x", "esd", r = 3)
outliers_test(df, "x", "dixon")
outliers_test(df, "x", "iqr", coef = 2)

Plot Arrhenius stability analysis

Description

Plot Arrhenius stability analysis

Usage

## S3 method for class 'arrhenius'
plot(x, ...)

Arguments

Value

A named list of ggplot objects, invisibly. The kinetic-fit and Arrhenius-regression plots are also drawn.

Examples

d <- data.frame(
  temperature = rep(c(25, 30, 37), each = 3),
  time = rep(c(0, 1, 3), 3),
  value = c(100, 98, 95, 100, 96, 90, 100, 94, 85)
)
fit <- arrhenius(d, "temperature", "time", "value",
                 target_temp = 5, limit = 10)
plot(fit)

Plot C5 and C95 results

Description

Plot C5 and C95 results

Usage

## S3 method for class 'c5_c95'
plot(x, type = c("probability", "profile"), ...)

Arguments

Value

A ggplot object.

Examples

set.seed(95)
d <- data.frame(level = rep(1:6, each = 12),
                y = stats::rnorm(72, rep(seq(36, 44, length.out = 6), each = 12), 1))
z <- c5_c95(d, "y", "level", cutoff = 40, model.no = 1)
plot(z, type = "probability")

Plot a commutability assessment

Description

Plot a commutability assessment

Usage

## S3 method for class 'commutability_result'
plot(x, method = NULL, ...)

Arguments

Value

A ggplot object, invisibly.

Examples

clinical <- expand.grid(
  sample = paste0("CS", 1:8), procedure = c("MP1", "MP2"),
  replicate = 1:3, stringsAsFactors = FALSE
)
level <- setNames(seq(10, 80, length.out = 8), paste0("CS", 1:8))
clinical$result <- level[clinical$sample] +
  ifelse(clinical$procedure == "MP2", 1, 0) +
  rep(c(-0.2, 0, 0.2), each = 16)
clinical$material_type <- "clinical"
clinical$position <- NA_integer_
rm <- expand.grid(
  sample = "RM1", procedure = c("MP1", "MP2"),
  position = 1:3, replicate = 1:3, stringsAsFactors = FALSE
)
rm$result <- 45 + ifelse(rm$procedure == "MP2", 1, 0) +
  rep(c(-0.1, 0, 0.1), each = 6)
rm$material_type <- "rm"
columns <- c("sample", "material_type", "procedure", "result",
             "position", "replicate")
example_data <- rbind(clinical[columns], rm[columns])
assessment <- commutability(
  example_data, "sample", "material_type", "procedure", "result"
)
plot(assessment)

Plot dilution-recovery results

Description

Plot dilution-recovery results

Usage

## S3 method for class 'dilution_recovery'
plot(x, type = c("recovery", "summary", "observed"), ...)

Arguments

Value

A ggplot object, invisibly.

Examples

dilution_data <- data.frame(
  sample = rep(c("S1", "S2"), each = 6),
  factor = rep(rep(c(1, 2, 4), each = 2), 2),
  result = c(99, 101, 49, 51, 24, 26,
             199, 201, 99, 101, 49, 51)
)
dilution <- dilution_recovery(
  dilution_data, "sample", "result", "factor"
)
plot(dilution)

Plot an equation fit

Description

Plot an equation fit

Usage

## S3 method for class 'fit_equation'
plot(x, interval = "confidence", level = 0.95, frame = FALSE, ...)

Arguments

Value

x, invisibly.

Examples

df4plc <- data.frame(
  conc = c(0.1, 0.3, 1, 3, 10, 30, 100),
  resp = c(12, 25, 48, 72, 88, 96, 99)
)
f <- fit_equation("E07", df4plc, "conc", "resp")
plot(f)
plot(f, interval = "prediction", frame = TRUE)

Plot a high-dose hook-effect experiment

Description

Plot a high-dose hook-effect experiment

Usage

## S3 method for class 'hook_effect'
plot(x, ...)

Arguments

Value

A ggplot object, invisibly.

Examples

hook_data <- data.frame(
  series = rep("S1", 8),
  concentration = rep(c(10, 20, 40, 80, 160, 320, 640, 1280), each = 2),
  response = c(10, 11, 20, 21, 39, 40, 70, 71,
               95, 94, 80, 79, 55, 54, 30, 29)
)
hook <- hook_effect(hook_data, "concentration", "response",
                    series = "series", uloq = 320)
plot(hook)

Plot interference dose-response results

Description

Plot interference dose-response results

Usage

## S3 method for class 'interference_dose_response'
plot(x, type = c("response", "absolute", "percent"), ...)

Arguments

Value

A ggplot object.

Examples

d <- expand.grid(conc = c(0, 10, 20), rep = 1:5)
d$result <- 100 + 0.5 * d$conc
plot(interference_dose_response(d, "result", "conc"), type = "percent")

Plot paired interference estimates

Description

Plot paired interference estimates

Usage

## S3 method for class 'interference_paired'
plot(x, ...)

Arguments

Value

A ggplot object.

Examples

d <- data.frame(group = rep(c("control", "test"), each = 5),
                result = c(10, 10.1, 9.9, 10.2, 9.8, 11, 11.1, 10.9, 11.2, 10.8))
plot(interference_paired(d, "result", "ep07", condition = "group"))

Plot patient-specimen interference results

Description

Plot patient-specimen interference results

Usage

## S3 method for class 'interference_patient'
plot(x, type = c("difference", "interferent"), interferent = NULL, ...)

Arguments

Value

A ggplot object.

Examples

d <- expand.grid(id = paste0("S", 1:8), method = c("test", "ref"), rep = 1:2)
d$group <- ifelse(as.integer(sub("S", "", d$id)) <= 4, "control", "selected")
d$bilirubin <- 2 * (as.integer(sub("S", "", d$id)) - 1)
d$result <- 20 + as.integer(sub("S", "", d$id)) +
  ifelse(d$method == "test", 0.1 * d$bilirubin, 0)
fit <- interference_patient(d, "id", "group", "method", "result",
  "selected", "control", "test", "ref", "bilirubin")
plot(fit, type = "difference")

Plot a linearity analysis

Description

Plot a linearity analysis

Usage

## S3 method for class 'linearity'
plot(
  x,
  type = c("fit", "replicates", "precision", "residual", "difference"),
  ...
)

Arguments

Value

A ggplot object.

Examples

set.seed(6)
d <- data.frame(level = rep(1:5, each = 3), x = rep(1:5, each = 3))
d$y <- 1 + 2 * d$x + stats::rnorm(nrow(d), sd = 0.2)
fit <- linearity(d, "level", "y", "x")
plot(fit, type = "fit")
plot(fit, type = "residual")

Plot (S3 method) – unified plotting interface

Description

Draw one of four plot types for an mcr object.

Usage

## S3 method for class 'mcr'
plot(
  x,
  type = c("scatter", "regression", "bland_altman", "bias"),
  interval = c("none", "confidence", "prediction", "both"),
  conf.level = NULL,
  mdl = NULL,
  ...
)

Arguments

Value

Invisible ggplot object

Examples

  obj <- mcr(ivd_mcr_example, "sid", "test", "ref")
  plot(obj)
  plot(obj, type = "scatter")
  obj <- regression(obj)
  plot(obj, type = "regression", interval = "confidence")
  obj <- bland_altman(obj)
  plot(obj, type = "bland_altman")
  obj <- bias(obj, mdl = c(200, 400))
  plot(obj, type = "bias")

Plot normality diagnostics

Description

Plot normality diagnostics

Usage

## S3 method for class 'normal_test'
plot(x, type = c("qq", "his"), ...)

Arguments

Value

The unchanged x, invisibly; requested plots are drawn as a side effect. For fewer than three observations no plot is drawn.

Examples

d <- data.frame(value = rnorm(30))
nt <- normal_test(d, "value")
plot(nt, type = c("qq", "his"))

S3 plot method

Description

Draw one of five plot types for a precision object. dot (run-order scatter plot) and his (histogram + N(0,1) density) require no prior analysis; var (variance component bar chart) requires variance() to be run first; qq (normal Q-Q plot) requires normal() first; profile (precision profile) requires profile() first.

Usage

## S3 method for class 'precision'
plot(x, type = c("dot", "his", "var", "qq", "profile"), ...)

Arguments

Value

Invisible ggplot object

Examples

  data(VCAdata1, package = "VCA")
  obj <- precision(VCAdata1, y ~ day/run, by = "sample")
  plot(obj, type = "dot")
  plot(obj, type = "his")

Plot a Levey-Jennings QC chart

Description

Plot a Levey-Jennings QC chart

Usage

## S3 method for class 'qc_chart'
plot(x, ...)

Arguments

Value

The unchanged x, invisibly; the chart is drawn as a side effect.

Examples

set.seed(42)
d <- data.frame(value = rnorm(30, 100, 5), run = 1:30)
plot(qc_chart(d, "value", run = "run"))

Plot reference interval results

Description

One panel per method: jitter strip chart overlaid with reference interval limits, endpoints, and CI error bars.

Usage

## S3 method for class 'reference_interval'
plot(x, ...)

Arguments

Value

A ggplot object, invisibly. The plot is also drawn as a side effect.

Examples

set.seed(11)
d <- data.frame(value = rnorm(120, 100, 15))
ri <- reference_interval(d, "value")
plot(ri)

Plot ROC or cutoff-dependent diagnostic metrics

Description

Draw ROC curves for evaluation columns and/or MLR model predictions, or draw sensitivity and specificity as functions of the cutoff for one evaluation column. Optionally mark optimal cutoff points (requires cutoff.roc() first).

Usage

## S3 method for class 'roc'
plot(x, curves = NULL, cutpoint = NULL, type = c("roc", "cutoff"), ...)

Arguments

Value

invisible ggplot object

Examples

  obj <- roc(ivd_roc_example, cols = c("x1", "x2"), reference = "ref")
  obj <- auc(obj)
  plot(obj)
  plot(obj, curves = "x1")
  # AUC calculation is not needed for the cutoff-dependent metric plot.
  plot(roc(ivd_roc_example, cols = "x1", reference = "ref"),
       curves = "x1", type = "cutoff")
  obj <- auc(obj); obj <- cutoff(obj); obj <- mlr(obj)
  plot(obj, curves = "all", cutpoint = "Youden")
  plot(obj, curves = "x1", type = "cutoff", cutpoint = "Youden")

Plot an analytical-sensitivity analysis

Description

Plot an analytical-sensitivity analysis

Usage

## S3 method for class 'sensitivity'
plot(x, ...)

Arguments

Value

A ggplot object.

Examples

d <- data.frame(sample = rep("blank", 10), result = seq(0.1, 1, length.out = 10))
fit <- sensitivity_lob(d, "result", "sample")
plot(fit)

Plot spike-recovery results

Description

Plot spike-recovery results

Usage

## S3 method for class 'spike_recovery'
plot(x, ...)

Arguments

Value

A ggplot object, invisibly.

Examples

spike_data <- data.frame(
  sample = rep(c("S1", "S2"), each = 4),
  condition = rep(rep(c("spiked", "control"), each = 2), 2),
  result = c(20.1, 19.9, 10.0, 10.2, 30.2, 29.8, 20.1, 19.9)
)
recovery <- spike_recovery(
  spike_data, "sample", "result", "condition",
  analyte_level = "spiked", solvent_level = "control",
  added_concentration = 10
)
plot(recovery)

Plot stability bias across storage conditions

Description

Plot stability bias across storage conditions

Usage

## S3 method for class 'stability_bias'
plot(x, ...)

Arguments

Value

A ggplot object, invisibly; the plot is also drawn.

Examples

d <- data.frame(
  condition = rep("2-8C", each = 8),
  time = rep(c(0, 1, 3, 7), 2),
  value = c(100, 99.8, 99.4, 98.9, 100.2, 99.9, 99.5, 98.8)
)
plot(stability_bias(d, "condition", "time", "value"))

Plot a stability regression

Description

Plot a stability regression

Usage

## S3 method for class 'stability_regression'
plot(x, ...)

Arguments

Value

A ggplot object, invisibly; the plot is also drawn.

Examples

d <- data.frame(
  condition = rep("2-8C", each = 8),
  time = rep(c(0, 1, 3, 7), 2),
  value = c(100, 99.8, 99.4, 98.9, 100.2, 99.9, 99.5, 98.8)
)
fit <- stability_regression(d, "time", "value", "condition")
plot(fit)

Plot stability limit-time predictions

Description

Plot stability limit-time predictions

Usage

## S3 method for class 'stability_time'
plot(x, ...)

Arguments

Value

A ggplot object, invisibly; the plot is also drawn.

Examples

d <- data.frame(
  condition = rep("2-8C", each = 8),
  time = rep(c(0, 1, 3, 7), 2),
  value = c(100, 99.8, 99.4, 98.9, 100.2, 99.9, 99.5, 98.8)
)
fit <- stability_regression(d, "time", "value", "condition")
limit_time <- stability_time(fit, limit = 5)
plot(limit_time)

Plot an uncertainty budget

Description

Plot an uncertainty budget

Usage

## S3 method for class 'uncertainty_budget'
plot(x, ...)

Arguments

Value

A ggplot object, invisibly.

Examples

repeatability <- uncertainty_type_a(
  c(9.9, 10.0, 10.1, 10.0), "repeatability", quantity = "mean"
)
calibrator <- uncertainty_type_b(
  "calibrator", estimate = 10, uncertainty = 0.2, k = 2,
  distribution = "normal"
)
budget <- uncertainty_combine(repeatability, calibrator, value = 10)
plot(budget)

Plot propagated uncertainty

Description

Plot propagated uncertainty

Usage

## S3 method for class 'uncertainty_propagation'
plot(x, type = c("distribution", "contribution"), ...)

Arguments

Value

A ggplot object, invisibly.

Examples

numerator <- uncertainty_type_b(
  "numerator", estimate = 10, uncertainty = 0.2, k = 2,
  distribution = "normal"
)
denominator <- uncertainty_type_b(
  "denominator", estimate = 2, uncertainty = 0.04, k = 2,
  distribution = "normal"
)
propagated <- uncertainty_propagate(
  function(numerator, denominator) numerator / denominator,
  list(numerator, denominator), method = "delta"
)
plot(propagated, type = "contribution")

Plot cumulative uncertainty through a traceability chain

Description

Plot cumulative uncertainty through a traceability chain

Usage

## S3 method for class 'uncertainty_traceability'
plot(x, ...)

Arguments

Value

A ggplot object, invisibly.

Examples

calibration <- uncertainty_type_b(
  "calibration", estimate = 100, uncertainty = 1, k = 2,
  distribution = "normal"
)
transport <- uncertainty_type_b(
  "transport", estimate = 100, lower = 99.5, upper = 100.5,
  distribution = "rectangular"
)
chain <- uncertainty_traceability(
  list(calibration = list(calibration), transport = list(transport)),
  value = 100
)
plot(chain)

Plot a Youden analysis

Description

Plot a Youden analysis

Usage

## S3 method for class 'youden_plot'
plot(x, ...)

Arguments

Value

The unchanged x, invisibly; the plot is drawn as a side effect.

Examples

set.seed(42)
d <- data.frame(m1 = rnorm(30, 100, 5), m2 = rnorm(30, 98, 5))
plot(youden_plot(d, "m1", "m2"))

Precision variance component analysis

Description

For each sample (grouped by by), estimate variance components via VCA::anovaVCA, compute SD / CV and their Satterthwaite confidence intervals.

Usage

precision(data, form, by = NULL, NegVC = FALSE, ...)

Arguments

Value

Returns an object of class precision. Use summary, profile, etc. to view results.

Examples

  data(VCAdata1, package = "VCA")
  precision(VCAdata1, y ~ day/run, by = "sample")

  # More complex: deeply nested factors with more sample groups
  data(realData, package = "VCA")
  precision(realData, y ~ lot/calibration/day/run, by = "PID")

Predict fitted values or inverse-predict (from y to x)

Description

Predict fitted values or inverse-predict (from y to x)

Usage

## S3 method for class 'fit_equation'
predict(
  object,
  newdata = NULL,
  x = NULL,
  y = NULL,
  inverse = FALSE,
  interval = NULL,
  level = 0.95,
  ...
)

Arguments

Value

data.frame with x, y, and (if requested) confidence/prediction interval columns

Examples

df4plc <- data.frame(
  conc = c(0.1, 0.3, 1, 3, 10, 30, 100),
  resp = c(12, 25, 48, 72, 88, 96, 99)
)
f <- fit_equation("E07", df4plc, "conc", "resp")

# Fitted values
predict(f)

# Confidence interval
predict(f, interval = "confidence")

# Prediction interval
predict(f, interval = "prediction")

# Inverse prediction: estimate x for y=50
predict(f, newdata = data.frame(resp = 50), inverse = TRUE)

Predict from an interference dose-response analysis

Description

Predict from an interference dose-response analysis

Usage

## S3 method for class 'interference_dose_response'
predict(
  object,
  newdata = NULL,
  target = NULL,
  metric = c("result", "absolute", "percent"),
  stratum = NULL,
  ...
)

Arguments

Value

A data frame of predictions or inverse interpolations.

Examples

d <- expand.grid(conc = c(0, 10, 20, 30, 40), rep = 1:5)
d$result <- 100 + 0.5 * d$conc
fit <- interference_dose_response(d, "result", "conc")
predict(fit, newdata = c(5, 15), metric = "percent")
predict(fit, target = 10, metric = "percent")

Predict (S3 method) – predict based on regression results

Description

Predict values based on stored regression results.

Usage

## S3 method for class 'mcr'
predict(
  object,
  newdata = NULL,
  reference = NULL,
  candidate = NULL,
  inverse = FALSE,
  interval = c("none", "confidence", "prediction", "both"),
  conf.level = 0.95,
  ci_method = c("auto", "analytic", "jackknife", "rank", "bootstrap_percentile",
    "bootstrap_standard"),
  bootstrap_replicates = 5000L,
  seed = NULL,
  ...
)

Arguments

Value

data.frame containing predicted values and (if requested) interval bounds

Examples

  obj <- mcr(ivd_mcr_example, "sid", "test", "ref")
  obj <- regression(obj, method = "ols")
  predict(obj, reference = c(40, 50, 60))
  predict(obj, reference = c(40, 50, 60), interval = "both")
  predict(obj, newdata = data.frame(xx = c(40, 50)), reference = "xx")
  predict(obj, inverse = TRUE)  # Inverse prediction on original data

Predict (S3 method) – predict from multivariate logistic regression

Description

Predict positive-class probability for new data using a stored MLR model. Returns the input predictor columns together with predicted probability, standard error, confidence interval, and 0/1 classification.

Usage

## S3 method for class 'roc'
predict(object, mlr = NULL, newdata = NULL, column_map = NULL, ...)

Arguments

Value

data.frame containing the predictor columns used in the model, plus:

Examples

  df <- data.frame(sid = 1:100,
                   x1  = c(rnorm(50, 10, 2), rnorm(50, 12, 2)),
                   x2  = rnorm(100, 5, 1),
                   ref = rep(c(0, 1), each = 50))
  obj <- roc(df, cols = c("x1", "x2"), reference = "ref")
  obj <- auc(obj)
  obj <- mlr(obj)
  predict(obj)
  predict(obj, newdata = df[1:5, ])
  ## column_map example: rename x1 to new_x1 in newdata
  new_df <- df[1:5, c("x2", "ref")]
  new_df$new_x1 <- df$x1[1:5]
  predict(obj, newdata = new_df,
          column_map = c(new_x1 = "x1", x2 = "x2"))

Print Arrhenius stability-analysis results

Description

Print Arrhenius stability-analysis results

Usage

## S3 method for class 'arrhenius'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

S3 print method: bottle_anova

Description

Print bottle ANOVA analysis results including descriptive statistics, ANOVA table, and assumption checks.

Usage

## S3 method for class 'bottle_anova'
print(x, ...)

Arguments

Value

Invisibly returns x.

Print C5 and C95 estimates

Description

Print C5 and C95 estimates

Usage

## S3 method for class 'c5_c95'
print(x, digits = 4L, ...)

Arguments

Value

Invisibly returns x.

Examples

set.seed(12)
d <- data.frame(level = rep(1:5, each = 10),
                y = stats::rnorm(50, rep(seq(37, 43, length.out = 5), each = 10), 1))
print(c5_c95(d, "y", "level", cutoff = 40, model.no = 1))

Print a commutability assessment

Description

Print a commutability assessment

Usage

## S3 method for class 'commutability_result'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print a multiple-equation comparison

Description

Print a multiple-equation comparison

Usage

## S3 method for class 'compare_equation'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print a fixed-cutpoint data split

Description

Print a fixed-cutpoint data split

Usage

## S3 method for class 'cutpoint_split'
print(x, ...)

Arguments

Value

The input object, invisibly.

S3 print method for describe_table

Description

S3 print method for describe_table

Usage

## S3 method for class 'describe_table'
print(x, ...)

Arguments

Value

Invisible describe_table object

Examples

  df <- data.frame(
    new    = c("positive","positive","negative","negative","positive","negative"),
    gold   = c("positive","negative","positive","negative","positive","positive"),
    age    = c(45, 52, 38, 61, 47, 55),
    gender = c("M","F","F","M","M","F"),
    id     = c("S001","S002","S003","S004","S005","S006")
  )
  tab <- raw_to_table(df, "new", "gold", id = "id")
  tab <- describe(tab)

S3 print method: format diagnostics output (diagnostic performance metrics)

Description

Prints four-fold table frequencies, sensitivity, specificity, PPV, NPV, accuracy, likelihood ratios, etc.

Usage

## S3 method for class 'diagnostics'
print(x, digits = 3, ...)

Arguments

Value

Invisible diagnostics object

Examples

  df <- data.frame(
    new    = c("positive","positive","negative","negative","positive","negative"),
    gold   = c("positive","negative","positive","negative","positive","positive"),
    age    = c(45, 52, 38, 61, 47, 55),
    gender = c("M","F","F","M","M","F"),
    id     = c("S001","S002","S003","S004","S005","S006")
  )
  tab <- raw_to_table(df, "new", "gold", positive = "positive", id = "id")
  tab <- diagnostics(tab)

Print dilution-recovery results

Description

Print dilution-recovery results

Usage

## S3 method for class 'dilution_recovery'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print a categorical data split

Description

Print a categorical data split

Usage

## S3 method for class 'factor_split'
print(x, ...)

Arguments

Value

The input object, invisibly.

Print an equation fit

Description

Print an equation fit

Usage

## S3 method for class 'fit_equation'
print(x, ...)

Arguments

Value

x, invisibly.

S3 print method: format a fourfold_table object as a fourfold table

Description

Prints data source, total sample size, variable names, fourfold table (with margins), and analysis status.

Usage

## S3 method for class 'fourfold_table'
print(x, margin = TRUE, ...)

Arguments

Value

Invisible fourfold_table object

Examples

  df <- data.frame(
    new    = c("positive","positive","negative","negative","positive","negative"),
    gold   = c("positive","negative","positive","negative","positive","positive"),
    age    = c(45, 52, 38, 61, 47, 55),
    gender = c("M","F","F","M","M","F"),
    id     = c("S001","S002","S003","S004","S005","S006")
  )
  tab <- raw_to_table(df, "new", "gold", id = "id")
  print(tab)

Print high-dose hook-effect results

Description

Print high-dose hook-effect results

Usage

## S3 method for class 'hook_effect'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print interference dose-response results

Description

Print interference dose-response results

Usage

## S3 method for class 'interference_dose_response'
print(x, ...)

Arguments

Value

x, invisibly.

Examples

d <- expand.grid(conc = c(0, 10, 20), rep = 1:5)
d$result <- 100 + 0.5 * d$conc
print(interference_dose_response(d, "result", "conc"))

Print paired interference estimates

Description

Print paired interference estimates

Usage

## S3 method for class 'interference_paired'
print(x, ...)

Arguments

Value

x, invisibly.

Examples

d <- data.frame(group = rep(c("control", "test"), each = 5),
                result = c(10, 10.1, 9.9, 10.2, 9.8, 11, 11.1, 10.9, 11.2, 10.8))
print(interference_paired(d, "result", "ep07", condition = "group"))

Print patient-specimen interference results

Description

Print patient-specimen interference results

Usage

## S3 method for class 'interference_patient'
print(x, ...)

Arguments

Value

x, invisibly.

Examples

d <- expand.grid(id = paste0("S", 1:6), method = c("test", "ref"), rep = 1:2)
d$group <- ifelse(as.integer(sub("S", "", d$id)) <= 3, "control", "selected")
d$result <- 10 + as.integer(sub("S", "", d$id)) + (d$method == "test")
print(interference_patient(d, "id", "group", "method", "result",
  "selected", "control", "test", "ref"))

Print EP07 replicate-count results

Description

Print EP07 replicate-count results

Usage

## S3 method for class 'interference_replicates'
print(x, ...)

Arguments

Value

x, invisibly.

Examples

print(interference_replicates(5, 7))

S3 print method: format kappa_table output

Description

Prints Cohen's Kappa / PABAK coefficient, standard error, confidence interval, and agreement rates.

Usage

## S3 method for class 'kappa_table'
print(x, digits = 4, ...)

Arguments

Value

Invisible kappa_table object

Examples

  tab <- counts_to_table(tp = 85, fp = 3, tn = 90, fn = 2,
                         candidate = "new method", reference = "gold standard")
  tab <- kappa(tab)

Print a linearity analysis

Description

Print a linearity analysis

Usage

## S3 method for class 'linearity'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

S3 print method: format mcnemar_table output

Description

Prints McNemar test results, including discordant pair counts, test statistic, and p-value.

Usage

## S3 method for class 'mcnemar_table'
print(x, ...)

Arguments

Value

Invisible mcnemar_table object

Examples

  tab <- counts_to_table(tp = 45, fp = 12, tn = 80, fn = 5,
                         candidate = "new method", reference = "gold standard")
  tab <- mcnemar(tab)

S3 print method

Description

Print data overview and analysis slot status based on $print slot.

Usage

## S3 method for class 'mcr'
print(x, ...)

Arguments

Value

Invisible mcr object

Examples

  df <- data.frame(sid = 1:30,
                   test = rnorm(30, 50, 10),
                   ref  = rnorm(30, 50, 10))
  obj <- mcr(df, "sid", "test", "ref")
  print(obj)

Print method for bias results

Description

Print method for bias results

Usage

## S3 method for class 'mcr_bias'
print(x, ...)

Arguments

Value

The unchanged mcr_bias data frame, invisibly.

Print method for Bland-Altman results

Description

Print method for Bland-Altman results

Usage

## S3 method for class 'mcr_bland_altman'
print(x, ...)

Arguments

Value

The unchanged mcr_bland_altman object, invisibly.

Print method for correlation results

Description

Print method for correlation results

Usage

## S3 method for class 'mcr_correlation'
print(x, ...)

Arguments

Value

The unchanged mcr_correlation object, invisibly.

Print method for describe results

Description

Print method for describe results

Usage

## S3 method for class 'mcr_describe'
print(x, digits = NULL, ...)

Arguments

Value

The unchanged mcr_describe object, invisibly.

Print method for outlier results

Description

Print method for outlier results

Usage

## S3 method for class 'mcr_outlier'
print(x, ...)

Arguments

Value

The unchanged mcr_outlier object, invisibly.

Print method for regression results

Description

Print method for regression results

Usage

## S3 method for class 'mcr_regression'
print(x, ...)

Arguments

Value

The unchanged mcr_regression object, invisibly.

Print mean kinetic temperature results

Description

Print mean kinetic temperature results

Usage

## S3 method for class 'mkt'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print a normality test result

Description

Print a normality test result

Usage

## S3 method for class 'normal_test'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print an outlier detection result

Description

Print an outlier detection result

Usage

## S3 method for class 'outliers_test'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print a precision object overview

Description

Print the data overview of a precision object (data class, total rows, formula, grouping variables, sample count, factor levels, balance warnings) and the analysis slot status ([x] / [ ]).

Usage

## S3 method for class 'precision'
print(x, ...)

Arguments

Value

Invisible precision object

Print confidence intervals

Description

Prints confidence interval tables for SD and \ sample, component, estimate, lower, upper.

Usage

## S3 method for class 'precision_ci'
print(x, ...)

Arguments

Value

Invisibly returns x.

Print normality test results

Description

Prints the selected normality test result per sample, including the actual method, sample size, statistic name, statistic, and p-value.

Usage

## S3 method for class 'precision_normal'
print(x, ...)

Arguments

Value

Invisibly returns x.

Print outlier detection results

Description

Prints the outlier detection method, count of outliers found, and the detail table (row, sample, value, statistic, critical value).

Usage

## S3 method for class 'precision_outlier'
print(x, ...)

Arguments

Value

Invisibly returns x.

Print precision profile fits

Description

Prints the best Sadler model formula, AIC, and R-squared for each variance component.

Usage

## S3 method for class 'precision_profile'
print(x, ...)

Arguments

Value

Invisibly returns x.

Print variance component table

Description

Prints the variance component summary table with columns: sample, component, VC (variance component), \

Usage

## S3 method for class 'precision_vc'
print(x, ...)

Arguments

Value

Invisibly returns x.

Print a Levey-Jennings QC chart analysis

Description

Print a Levey-Jennings QC chart analysis

Usage

## S3 method for class 'qc_chart'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print reference-material bias results

Description

Print reference-material bias results

Usage

## S3 method for class 'reference_bias'
print(x, ...)

Arguments

Value

x, invisibly.

Examples

d <- data.frame(result = c(9.8, 10.0, 10.1, 9.9, 10.2, 10.0))
print(reference_bias(d, "result", reference = 10,
                     reference_uncertainty = 0.2))

Print reference interval analysis results

Description

Print reference interval analysis results

Usage

## S3 method for class 'reference_interval'
print(x, ...)

Arguments

Value

The unchanged reference_interval object, invisibly.

Print a ROC analysis object

Description

Print a ROC analysis object

Usage

## S3 method for class 'roc'
print(x, ...)

Arguments

Value

x, invisibly.

Print a paired AUC comparison

Description

Print a paired AUC comparison

Usage

## S3 method for class 'roc_auc_comparison'
print(x, ...)

Arguments

Value

The input object, invisibly.

Print method for roc AUC table

Description

Print method for roc AUC table

Usage

## S3 method for class 'roc_auc_table'
print(x, ...)

Arguments

Value

The unchanged roc_auc_table data frame, invisibly.

Print method for roc cutoff table

Description

Print method for roc cutoff table

Usage

## S3 method for class 'roc_cutoff_table'
print(x, ...)

Arguments

Value

The unchanged roc_cutoff_table data frame, invisibly.

Print method for roc describe results

Description

Print method for roc describe results

Usage

## S3 method for class 'roc_describe'
print(x, digits = NULL, ...)

Arguments

Value

The unchanged roc_describe object, invisibly.

Print a multivariate ROC model

Description

Print a multivariate ROC model

Usage

## S3 method for class 'roc_mlr'
print(x, ...)

Arguments

Value

x, invisibly.

Print Bland-Altman sample-size results

Description

Print Bland-Altman sample-size results

Usage

## S3 method for class 'sample_size_bland_altman'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print single-proportion target-value sample-size results

Description

Print single-proportion target-value sample-size results

Usage

## S3 method for class 'sample_size_proportion'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print single-proportion confidence-interval sample-size results

Description

Print single-proportion confidence-interval sample-size results

Usage

## S3 method for class 'sample_size_proportion_ci'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print an analytical-sensitivity analysis result

Description

Print an analytical-sensitivity analysis result

Usage

## S3 method for class 'sensitivity'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print an analytical-sensitivity study design

Description

Print an analytical-sensitivity study design

Usage

## S3 method for class 'sensitivity_design'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print spike-concentration calculations

Description

Print spike-concentration calculations

Usage

## S3 method for class 'spike_concentration'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print spike-recovery results

Description

Print spike-recovery results

Usage

## S3 method for class 'spike_recovery'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print stability-bias results

Description

Print stability-bias results

Usage

## S3 method for class 'stability_bias'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print a stability study time-point plan

Description

Print a stability study time-point plan

Usage

## S3 method for class 'stability_plan'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print stability-regression results

Description

Print stability-regression results

Usage

## S3 method for class 'stability_regression'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print stability limit-time predictions

Description

Print stability limit-time predictions

Usage

## S3 method for class 'stability_time'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

S3 print method: tukey

Description

Print Tukey HSD post-hoc test results including pairwise comparison table, significance markers, and compact letter display.

Usage

## S3 method for class 'tukey'
print(x, ...)

Arguments

Value

Invisibly returns x.

Print a combined uncertainty budget

Description

Print a combined uncertainty budget

Usage

## S3 method for class 'uncertainty_budget'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print characterization uncertainty results

Description

Print characterization uncertainty results

Usage

## S3 method for class 'uncertainty_characterization'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print an uncertainty component

Description

Print an uncertainty component

Usage

## S3 method for class 'uncertainty_component'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print between-unit homogeneity uncertainty results

Description

Print between-unit homogeneity uncertainty results

Usage

## S3 method for class 'uncertainty_homogeneity'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print IQC uncertainty results

Description

Print IQC uncertainty results

Usage

## S3 method for class 'uncertainty_iqc'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print propagated uncertainty results

Description

Print propagated uncertainty results

Usage

## S3 method for class 'uncertainty_propagation'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print stability uncertainty results

Description

Print stability uncertainty results

Usage

## S3 method for class 'uncertainty_stability'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print traceability-chain uncertainty results

Description

Print traceability-chain uncertainty results

Usage

## S3 method for class 'uncertainty_traceability'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

Print a Youden plot analysis

Description

Print a Youden plot analysis

Usage

## S3 method for class 'youden_plot'
print(x, ...)

Arguments

Value

The unchanged x, invisibly.

S3 precision profile method

Description

Fit precision (SD) vs. mean across samples using the Sadler model selection algorithm (10 candidate models) based on VFP::fit_vfp().

Usage

## S3 method for class 'precision'
profile(object, model.no = 1:10, ...)

Arguments

Value

Updated precision object with results stored in $profile

Examples

  data(VCAdata1, package = "VCA")
  obj <- precision(VCAdata1, y ~ day/run, by = "sample")
  obj <- variance(obj)
  obj <- profile(obj, model.no = 1)

Levey-Jennings QC chart + Westgard rule evaluation

Description

Draws a Levey-Jennings QC chart and detects out-of-control points based on selected Westgard rules. If target mean and SD are not provided, they are estimated from the data.

Usage

qc_chart(
  data,
  value,
  rules = "1-2s,1-3s,2-2s,R-4s,4-1s,10x",
  mean = NULL,
  sd = NULL,
  group = NULL,
  run = NULL
)

Arguments

Value

An object of class "qc_chart" with violation results and plot data. Use print() to show summary and plot() to draw the Levey-Jennings chart.

Examples

  set.seed(42)
  df <- data.frame(
    run   = 1:30,
    value = c(rnorm(27, 100, 5), 108, 115, 85)
  )
  res <- qc_chart(df, "value", rules = "1-2s,1-3s,10x")
  print(res)
  plot(res)

Core function: raw data -> frequency fourfold table (silent construction)

Description

Core function: raw data -> frequency fourfold table (silent construction)

Usage

raw_to_table(
  data,
  candidate,
  reference,
  id = NULL,
  na.rm = TRUE,
  positive = NULL
)

Arguments

Value

A fourfold_table object (list with S3 class "fourfold_table"), containing: - $freq_raw Raw frequencies (long format) - $table Fourfold matrix (with margins) - $n Total sample size - $print Print slot (descriptive metadata) - $candidate_levels Levels of the candidate method - $reference_levels Levels of the reference method Use print() directly to display the fourfold table.

Examples

  df <- data.frame(
    new    = c("positive","positive","negative","negative","positive","negative"),
    gold   = c("positive","negative","positive","negative","positive","positive"),
    age    = c(45, 52, 38, 61, 47, 55),
    gender = c("M","F","F","M","M","F"),
    id     = c("S001","S002","S003","S004","S005","S006")
  )
  result <- raw_to_table(df, "new", "gold", id = "id")

Estimate bias using reference materials

Description

Calculates measurement bias and its interval according to Section 5.3 of YY/T 1789.2-2021. For each reference-material level, the uncertainty of the measured mean is calculated as s / sqrt(n) and expanded by k. This expanded uncertainty is then combined with the reference-material uncertainty. The function reports results only and applies no acceptance criterion.

Usage

reference_bias(
  data,
  result,
  reference,
  reference_uncertainty,
  level = NULL,
  k = 2
)

Arguments

Value

An object of class "reference_bias". Its result component contains the sample size, mean, standard deviation, uncertainty, bias, relative bias, and interval for each reference-material level.

References

YY/T 1789.2-2021, Section 5.3 and Appendix A.

Examples

reference_data <- data.frame(
  level = rep(c("low", "high"), each = 6),
  result = c(9.8, 10.1, 10.0, 9.9, 10.2, 10.0,
             19.7, 20.1, 20.0, 19.9, 20.2, 20.1)
)
bias_result <- reference_bias(
  reference_data, result = "result",
  reference = c(10, 20),
  reference_uncertainty = c(0.2, 0.3),
  level = "level"
)
bias_result$result

Reference interval analysis

Description

Calculates reference intervals using the CLSI EP28-A3c simple nonparametric or biweight procedures, a normal/log-normal parametric procedure, or the retained R type-6 and Huber procedures.

Usage

reference_interval(
  data,
  col,
  interval = 0.95,
  ci = 0.95,
  method = "nonparametric",
  id = NULL,
  bootstrap_replicates = 1999L,
  seed = NULL
)

Arguments

Details

method = "nonparametric" uses ranks p * (n + 1), linear interpolation, and exact binomial order-statistic confidence intervals. method = "biweight" implements the EP28 Appendix B biweight location and scale calculation and percentile bootstrap confidence intervals. method = "type6" retains the previous R type-6 quantiles with normal-approximation rank intervals. method = "huber" retains the previous Huber M-estimator.

Value

An object of class reference_interval. Selected results are stored in correspondingly named elements: nonparametric, biweight, type6, parametric, and huber. Data-quality counts and cleaned values are also retained.

References

Clinical and Laboratory Standards Institute. Defining, Establishing, and Verifying Reference Intervals in the Clinical Laboratory; Approved Guideline—Third Edition. CLSI document EP28-A3c. Wayne, PA: CLSI; 2010.

Examples

set.seed(17)
d <- data.frame(id = seq_len(150), value = rnorm(150, 100, 15))
reference_interval(d, "value", id = "id")
reference_interval(d, "value", method = c("nonparametric", "parametric"))
reference_interval(d, "value", method = "all",
                   bootstrap_replicates = 199, seed = 28)

Regression analysis (S3 method)

Description

Perform regression analysis with the reference/comparative method on the X-axis and the candidate/test method on the Y-axis. Supports five methods: Ordinary Least Squares (OLS), Weighted Least Squares (WLS), Deming regression, Weighted Deming regression, and Passing-Bablok regression.

Usage

## S3 method for class 'mcr'
regression(
  x,
  method = c("ols", "wls", "deming", "wdeming", "pb"),
  conf.level = 0.95,
  lambda = 1,
  weights = NULL,
  variance_models = NULL,
  profile_components = c(reference = "total", candidate = "total"),
  ci_method = c("auto", "analytic", "jackknife", "rank", "bootstrap_percentile",
    "bootstrap_standard"),
  bootstrap_replicates = 5000L,
  seed = NULL,
  mdl = NULL,
  weight_iterations = 4L,
  epsilon = NULL,
  ...
)

Arguments

Value

Updated mcr object (invisible), results stored in $regression

Examples

  obj <- mcr(ivd_mcr_example, "sid", "test", "ref")
  regression(obj, method = "ols")
  regression(obj, method = "deming")

Summarize replicate measurements

Description

Groups data by x, computes mean, SD, and sample size for each group. When weights is specified, returns an additional weight column. Summary metadata (call, weighting method, column names) is stored as attributes for downstream reference.

Usage

replicate_to_mean(data, x, y, weights = NULL, na.rm = TRUE)

Arguments

Value

A data.frame with columns:

Attributes: call, weights (method name), x_col, y_col

Examples

# Basic summary
df <- data.frame(
  conc = rep(c(1, 2, 5, 10), each = 3),
  resp = c(3.1, 3.2, 2.9, 5.8, 6.1, 5.9,
           14.2, 14.5, 14.0, 28.1, 27.9, 28.3)
)
rep <- replicate_to_mean(df, "conc", "resp")
rep

# Inverse-variance weighting
rep_w <- replicate_to_mean(df, "conc", "resp", weights = "1/sd^2")
rep_w

Generate an EP05 precision summary table

Description

Generate an EP05 precision summary table

Usage

## S3 method for class 'precision'
report(
  x,
  day,
  run = NULL,
  site = NULL,
  table = c("simple", "detailed", "CI"),
  digits = 4L,
  ...
)

Arguments

Details

Interaction terms are matched independently of the textual order of their factors. For a multisite design, within-laboratory precision is formed from the within-site components (for example, site:day + error), whereas reproducibility includes the between-site component. With table = "CI", confidence intervals for combined components use the variance-component covariance matrix and a Satterthwaite approximation.

Value

A data.frame containing the requested report table.

Examples

if (requireNamespace("VCA", quietly = TRUE)) {
  data(VCAdata1, package = "VCA")
  p <- precision(VCAdata1, y ~ day/run, by = "sample")
  p <- variance(p)
  report(p, day = "day", run = "run", table = "simple")
}

Extract residuals from an equation fit

Description

Extract residuals from an equation fit

Usage

## S3 method for class 'fit_equation'
residuals(object, ...)

Arguments

Value

A numeric vector of residuals.

Examples

df <- data.frame(x = 1:5, y = c(2.1, 4.0, 6.2, 7.9, 10.1))
f <- fit_equation("E01", df, "x", "y")
residuals(f)

roc constructor

Description

Create an ROC analysis object.

Usage

roc(data, cols, reference, id = NULL, positive = NULL)

Arguments

Details

Supports multiple evaluation columns (quantitative) and a single reference column (binary 0/1, factor, or character with two levels). Automatically detects direction, reports missing values and ID duplicates.

Value

Returns an S3 object of class "roc"

Examples

  df <- data.frame(
    sid = 1:100,
    x1  = c(rnorm(50, 10, 2), rnorm(50, 12, 2)),
    ref = rep(c(0, 1), each = 50),
    age = 1:100,
    sex = rep(c("F", "M"), each = 50)
  )
  obj <- roc(df, cols = "x1", reference = "ref", id = "sid")
  print(obj)

Sample size for Bland-Altman agreement assessment

Description

Calculates the minimum number of measurement pairs required to attain a target power using the noncentral-t approximation of Lu et al. (2016).

Usage

sample_size_bland_altman(
  power = 0.8,
  mu,
  sd,
  delta,
  conf.level = 0.95,
  agree.level = 0.95,
  method = "lu",
  n.min = 3L,
  n.max = 1e+05
)

Arguments

Value

An object of class sample_size_bland_altman. The required sample size is available as x$n; the requested and attained powers are available as x$target_power and x$achieved_power.

References

Lu MJ, Zhong WH, Liu YX, Miao HZ, Li YC, Ji MH (2016). Sample size for assessing agreement between two methods of measurement by Bland-Altman method. International Journal of Biostatistics, 12(2). doi:10.1515/ijb-2015-0039

Examples

sample_size_bland_altman(
  power = 0.80, mu = 0.2, sd = 1, delta = 2.5
)

Sample size for a single-proportion target-value test

Description

Calculates the minimum sample size required to test a single proportion against a target value. The test statistic and the power-calculation method are selected independently. Regardless of the planning method, the returned result includes the exact binomial Type I error and power of the resulting discrete rejection region.

Usage

sample_size_proportion(
  p0,
  p1,
  alpha = 0.05,
  power = 0.8,
  alternative = c("greater", "less", "two.sided"),
  test = c("exact", "score", "score-cc", "wald", "wald-cc", "agresti-coull", "jeffreys"),
  power_method = c("binomial", "normal"),
  n.max = 1e+06
)

Arguments

Details

⁠test = "score", power_method = "normal"⁠ uses the null variance for the alpha term and the alternative variance for the beta term. This is the normal-approximation procedure used by the CMDE target-value formula. With an infinite population, test = "exact" and test = "score" have the same normal-approximation formula, matching PASS. The test choice still determines the discrete rejection region used for the reported actual alpha and achieved power.

Value

An object of class sample_size_proportion. The required sample size is available as x$n. The object also contains target and planning power, exact achieved power, actual alpha, and critical success counts.

Examples

# Exact binomial design
sample_size_proportion(
  p0 = 0.2, p1 = 0.35, alpha = 0.05, power = 0.8,
  alternative = "greater"
)

# CMDE continuous normal approximation
sample_size_proportion(
  p0 = 0.85, p1 = 0.90, alpha = 0.05, power = 0.8,
  alternative = "two.sided", test = "score", power_method = "normal"
)

Sample size for a single-proportion confidence interval

Description

Calculates the minimum sample size whose two-sided confidence interval has total width no greater than twice half_width, or whose one-sided limit is no farther than half_width from the anticipated sample proportion.

Usage

sample_size_proportion_ci(
  p,
  half_width,
  conf.level = 0.95,
  interval = c("two.sided", "lower", "upper"),
  method = c("wilson", "wald", "wald-cc", "agresti-coull", "jeffreys", "wilson-cc",
    "clopper-pearson"),
  n.max = 1e+06
)

Arguments

Value

An object of class sample_size_proportion_ci. The required sample size is available as x$n. The returned object also contains the achieved half-width and the exact binomial coverage at the assumed value of p.

Examples

sample_size_proportion_ci(p = 0.3, half_width = 0.05)
sample_size_proportion_ci(
  p = 0.3, half_width = 0.03, method = "clopper-pearson"
)

Summarize an analytical-sensitivity study design

Description

Returns minimum CLSI EP17-A2 design requirements and, optionally, counts observed in a supplied data set. Differences are descriptive only; the function does not issue a pass/fail conclusion.

Usage

sensitivity_design(
  method = c("classical", "profile", "probit", "loq", "verification"),
  data = NULL,
  result = NULL,
  sample = NULL,
  lot = NULL,
  day = NULL,
  concentration = NULL,
  limit = c("lob", "lod", "loq")
)

Arguments

Value

An object of class sensitivity_design.

References

CLSI. Evaluation of Detection Capability for Clinical Laboratory Measurement Procedures. EP17-A2, 2nd ed. 2012.

Examples

sensitivity_design(c("classical", "loq"))

design_data <- data.frame(
  result = 1:8, sample = rep(c("S1", "S2"), each = 4),
  lot = rep(c("A", "B"), 4), day = rep(1:2, each = 4)
)
sensitivity_design("verification", data = design_data,
                   result = "result", sample = "sample",
                   lot = "lot", day = "day", limit = "lod")

Establish a limit of blank

Description

Establishes a LoB from replicate-level blank results using the CLSI EP17-A2 nonparametric, parametric, or zero-LoB procedure. With two or three reagent lots, lots are analyzed separately and the maximum estimate is reported; four or more lots are combined by default.

Usage

sensitivity_lob(
  data,
  result,
  sample,
  lot = NULL,
  method = c("nonparametric", "parametric", "zero"),
  alpha = 0.05,
  lot_rule = c("ep17", "separate", "combined"),
  exclude = NULL
)

Arguments

Value

An object of class sensitivity_lob.

References

CLSI. Evaluation of Detection Capability for Clinical Laboratory Measurement Procedures. EP17-A2, 2nd ed. 2012.

Examples

blank <- data.frame(
  sample = rep(c("blank_1", "blank_2"), each = 5),
  result = c(0.02, 0.04, 0.03, 0.05, 0.01,
             0.03, 0.06, 0.04, 0.02, 0.05)
)
lob <- sensitivity_lob(blank, result = "result", sample = "sample")
lob$estimate

Establish a limit of detection

Description

Establishes LoD using the classical, nonparametric trial, precision-profile, or probit approach described in CLSI EP17-A2.

Usage

sensitivity_lod(
  data,
  result = NULL,
  sample = NULL,
  lob,
  lot = NULL,
  concentration = NULL,
  detected = NULL,
  method = c("classical", "nonparametric", "profile", "probit"),
  beta = 0.05,
  profile_model = c("linear", "quadratic", "sadler", "vfp"),
  concentration_scale = c("log10", "identity"),
  total = NULL,
  strata = NULL,
  model_no = 1:9,
  lot_rule = c("ep17", "separate", "combined"),
  exclude = NULL
)

Arguments

Value

An object of class sensitivity_lod.

References

CLSI. Evaluation of Detection Capability for Clinical Laboratory Measurement Procedures. EP17-A2, 2nd ed. 2012.

Examples

low <- data.frame(
  sample = rep(c("low_1", "low_2"), each = 5),
  result = c(0.22, 0.26, 0.24, 0.25, 0.23,
             0.31, 0.29, 0.33, 0.30, 0.32)
)
lod <- sensitivity_lod(low, "result", "sample", lob = 0.10,
                       method = "classical")
lod$estimate

Establish a limit of quantitation

Description

Establishes LoQ from Westgard total error, RMS error, a precision goal, or a user-supplied accuracy function. The original candidate and the LoD- constrained reported LoQ are retained separately.

Usage

sensitivity_loq(
  data,
  result,
  sample,
  assigned,
  lot = NULL,
  method = c("westgard", "rms", "precision", "custom"),
  goal,
  goal_scale = c("relative", "absolute"),
  lod = NULL,
  k = 2,
  accuracy_function = NULL,
  profile = FALSE,
  profile_model = c("linear", "quadratic"),
  lot_rule = c("ep17", "separate", "combined"),
  exclude = NULL,
  ...
)

Arguments

Value

An object of class sensitivity_loq.

References

CLSI. Evaluation of Detection Capability for Clinical Laboratory Measurement Procedures. EP17-A2, 2nd ed. 2012.

Examples

loq_data <- data.frame(
  sample = rep(paste0("L", 1:3), each = 4),
  assigned = rep(c(0.5, 1, 2), each = 4),
  result = c(0.42, 0.48, 0.51, 0.46,
             0.91, 1.02, 0.96, 1.05,
             1.91, 2.04, 1.98, 2.08)
)
loq <- sensitivity_loq(
  loq_data, "result", "sample", assigned = "assigned",
  method = "precision", goal = 0.15, goal_scale = "relative"
)
loq$summary

Verify an LoB, LoD, or LoQ claim

Description

Calculates the proportion of verification results consistent with a stated claim, its one-sided Wilson score limits, and the CLSI EP17-A2 Table 1 boundary. No pass/fail field or regulatory conclusion is produced.

Usage

sensitivity_verify(
  data,
  result,
  sample,
  limit = c("lob", "lod", "loq"),
  claim,
  lob = NULL,
  assigned = NULL,
  allowable_error = NULL,
  error_scale = c("relative", "absolute"),
  alpha = 0.05,
  exclude = NULL
)

Arguments

Value

An object of class sensitivity_verification.

References

CLSI. Evaluation of Detection Capability for Clinical Laboratory Measurement Procedures. EP17-A2, 2nd ed. 2012.

Examples

verification <- data.frame(
  sample = rep(c("blank_1", "blank_2"), each = 10),
  result = c(rep(0.03, 9), 0.07, rep(0.04, 9), 0.08)
)
verified <- sensitivity_verify(
  verification, "result", "sample", limit = "lob", claim = 0.05
)
verified$result

Calculate the concentration of a spiked sample

Description

Applies the mass-balance equation for mixing a sample and a spiking material. Arguments are vectorized using ordinary R recycling.

Usage

spike_concentration(
  sample_concentration,
  sample_volume,
  stock_concentration,
  spike_volume
)

Arguments

Value

A data frame containing total volume, spike fraction, concentration added by the spike, and final calculated concentration.

References

CLSI. Establishing and Verifying an Extended Measuring Interval Through Specimen Dilution and Spiking. EP34, 1st ed. 2018.

Examples

spike_concentration(
  sample_concentration = 10, sample_volume = 0.95,
  stock_concentration = 210, spike_volume = 0.05
)

Calculate spike recovery against a solvent control

Description

Calculates the difference between analyte-spiked and equal-volume solvent-control means and expresses that difference as a percentage of the known added concentration. No acceptance criterion is applied.

Usage

spike_recovery(
  data,
  sample,
  result,
  condition,
  analyte_level,
  solvent_level,
  added_concentration = NULL,
  sample_volume = NULL,
  stock_concentration = NULL,
  spike_volume = NULL,
  conf.level = 0.95
)

Arguments

Value

An object of class spike_recovery.

References

CLSI. Establishing and Verifying an Extended Measuring Interval Through Specimen Dilution and Spiking. EP34, 1st ed. 2018.

Examples

spike_data <- data.frame(
  sample = rep(c("S1", "S2"), each = 4),
  condition = rep(rep(c("spiked", "control"), each = 2), 2),
  result = c(20.1, 19.9, 10.0, 10.2, 30.2, 29.8, 20.1, 19.9)
)
recovery <- spike_recovery(
  spike_data, "sample", "result", "condition",
  analyte_level = "spiked", solvent_level = "control",
  added_concentration = 10
)
recovery$results

Split a data frame at fixed numeric cutpoints

Description

Split an input data frame into a named list of data frames before analysis. Cutpoints are applied to one numeric variable using left-closed, right-open intervals. For example, cutpoints c(100, 500) create ⁠[-Inf, 100)⁠, ⁠[100, 500)⁠, and ⁠[500, Inf)⁠.

Usage

split_by_cutpoints(
  data,
  variable,
  cutpoints,
  labels = NULL,
  na_action = c("error", "drop"),
  drop_empty = FALSE
)

Arguments

Value

A named list of data frames with class cutpoint_split. The segment_info attribute records interval bounds, labels, and row counts.

Examples

parts <- split_by_cutpoints(
  ivd_mcr_example, variable = "ref", cutpoints = c(18, 26)
)
parts
parts$segment_1
fits <- lapply(parts, function(d) mcr(d, "sid", "test", "ref"))

Split a data frame by the levels of one categorical variable

Description

Split an input data frame into a named list of data frames before analysis. Factor levels retain their declared order; character and logical values use first-appearance order. The splitting column is retained in every returned data frame.

Usage

split_by_factors(
  data,
  variable,
  na_action = c("drop", "error"),
  drop_empty = TRUE
)

Arguments

Value

A named list of data frames with class factor_split. List names are the original factor levels or categorical values. The group_info attribute records their order and row counts.

Examples

d <- data.frame(
  batch = factor(c("Batch-A", "Batch-A", "Batch-B")),
  result = c(10.1, 9.9, 10.4)
)
parts <- split_by_factors(d, variable = "batch")
parts
parts[["Batch-A"]]
means <- lapply(parts, function(z) mean(z$result))

stability_bias — Node-wise bias analysis across storage conditions

Description

Each storage condition is baseline-corrected to its own first time point. Absolute and relative biases are computed and compared across conditions at matched time points.

Usage

stability_bias(
  data,
  condition,
  time,
  value,
  reference_condition = NULL,
  limit = NULL,
  bias_type = c("relative", "absolute"),
  time_unit = "d"
)

Arguments

Value

An S3 object of class "stability_bias".

Examples

set.seed(42)
df <- data.frame(cond = rep(c("2-8C","25C"), each = 8),
                 t = rep(c(0,1,3,7), 4),
                 val = c(rnorm(4,100,2), rnorm(4,98,2),
                         rnorm(4,100,2), rnorm(4,92,3)))
stability_bias(df, "cond", "t", "val", limit = 5)

stability_plan — CLSI EP25 Appendix A time-point planning

Description

Based on the ratio of expected drift to allowable drift and the reproducibility variability, looks up the minimum number of time points required per regression series in the EP25 Appendix A tables.

Usage

stability_plan(
  allowable_drift,
  variability,
  replicates = 1L,
  expected_drift = NULL,
  power = c(0.8, 0.9),
  mode = c("simple", "components"),
  bias_type = c("relative", "absolute")
)

Arguments

Value

An S3 object of class "stability_plan".

Examples

stability_plan(4, 1.5, replicates = c(1,2,3))

comp <- data.frame(source = c("repeatability","lot","operator"),
                   variability = c(1.5,1.0,0.8),
                   levels = c(1,3,2))
stability_plan(4, comp, replicates = c(1,2,3), mode = "components")

stability_regression — CLSI EP25 regression-based stability assessment

Description

Performs simple linear regression on a single condition's measurements (self mode) or on paired biases between test and reference conditions (compare mode), with one-sided confidence limits to evaluate whether drift remains within allowable limits.

Usage

stability_regression(
  data,
  time,
  value,
  condition = NULL,
  mode = c("self", "compare"),
  test_condition = NULL,
  reference_condition = NULL,
  bias_type = c("relative", "absolute"),
  direction = c("auto", "decrease", "increase"),
  conf.level = 0.95,
  limit = NULL,
  time_unit = "d"
)

Arguments

Value

An S3 object of class "stability_regression".

Examples

set.seed(42)
df <- data.frame(cond = rep(c("2-8C","25C"), each = 8),
                 t = rep(c(0,1,3,7), 4),
                 val = c(rnorm(4,100,2), rnorm(4,98,2),
                         rnorm(4,100,2), rnorm(4,92,3)))
stability_regression(df, "t", "val", "cond", test_condition = "25C", limit = 5)

set.seed(42)
df <- data.frame(cond = rep(c("Ref","Test"), each = 8),
                 t = rep(c(0,1,3,7), 4),
                 val = c(rnorm(4,100,2), rnorm(4,99,2),
                         rnorm(4,100,2), rnorm(4,94,3)))
stability_regression(df, "t", "val", "cond", mode = "compare",
                     test_condition = "Test", reference_condition = "Ref", limit = 5)

stability_time — Predict time-to-limit from a stability regression

Description

Given a fitted regression model, computes the time at which the point estimate or one-sided confidence limit first reaches the allowable bias. May extrapolate beyond the observed time range.

Usage

stability_time(object, limit = NULL, max_time = NULL)

Arguments

Value

An S3 object of class "stability_time".

Examples

set.seed(42)
df <- data.frame(cond = rep(c("2-8C","25C"), each = 8),
                 t = rep(c(0,1,3,7), 4),
                 val = c(rnorm(4,100,2), rnorm(4,98,2),
                         rnorm(4,100,2), rnorm(4,92,3)))
r <- stability_regression(df, "t", "val", "cond", test_condition = "25C", limit = 5)
stability_time(r)

set.seed(42)
df <- data.frame(cond = rep(c("Ref","Test"), each = 8),
                 t = rep(c(0,1,3,7), 4),
                 val = c(rnorm(4,100,2), rnorm(4,99,2),
                         rnorm(4,100,2), rnorm(4,94,3)))
r2 <- stability_regression(df, "t", "val", "cond", mode = "compare",
                           test_condition = "Test", reference_condition = "Ref", limit = 5)
stability_time(r2, limit = 5)

Summarize a commutability assessment

Description

Summarize a commutability assessment

Usage

## S3 method for class 'commutability_result'
summary(object, ...)

Arguments

Value

The unchanged object, invisibly.

S3 summary method

Description

Summarizes all key analysis information already executed in a fourfold_table object, formatted following summary.mcr() pattern (mcr.R).

Usage

## S3 method for class 'fourfold_table'
summary(object, digits = 3, ...)

Arguments

Value

Invisible fourfold_table object

Examples

  tab <- counts_to_table(tp = 85, fp = 3, tn = 90, fn = 2,
                         candidate = "new method", reference = "gold standard")
  summary(tab)
  tab <- diagnostics(tab); tab <- kappa(tab); tab <- mcnemar(tab)
  summary(tab)

S3 summary method

Description

Summarize key information from all executed analyses in the mcr object (descriptive statistics, correlation analysis, regression analysis, outlier detection, Bland-Altman analysis, and bias analysis).

Usage

## S3 method for class 'mcr'
summary(object, ...)

Arguments

Value

Invisible mcr object

Examples

  df <- data.frame(sid = 1:30,
                   test = rnorm(30, 50, 10),
                   ref  = rnorm(30, 50, 10))
  obj <- mcr(df, "sid", "test", "ref")
  summary(obj)
  obj <- correlation(obj)
  obj <- regression(obj, method = "ols")
  obj <- bland_altman(obj)
  summary(obj)

S3 summary method

Description

Summarize all executed analyses in a precision object, including data description, analysis status checklist, outlier detection, normality tests, variance components, confidence intervals, and precision profile.

Usage

## S3 method for class 'precision'
summary(object, ...)

Arguments

Value

Invisible precision object

Examples

  data(VCAdata1, package = "VCA")
  obj <- precision(VCAdata1, y ~ day/run, by = "sample")
  summary(obj)

Summary (S3 method)

Description

Output key information from all analyses performed on the roc object.

Usage

## S3 method for class 'roc'
summary(object, ...)

Arguments

Value

invisible roc object

Examples

  obj <- roc(ivd_roc_example, cols = c("x1", "x2"), reference = "ref")
  summary(obj)
  obj <- auc(obj)
  obj <- cutoff(obj)
  obj <- mlr(obj)
  summary(obj)

Tukey HSD post-hoc test (S3 method)

Description

Perform Tukey HSD pairwise comparisons on an ANOVA result, with compact letter display.

Usage

## S3 method for class 'bottle_anova'
tukey(x, term = NULL, conf.level = NULL, ...)

Arguments

Value

An S3 object of class "tukey" with components:

Note: The original bottle_anova object's $tukey_results is also updated.

Examples

  obj <- bottle_anova(ivd_bottle_example, value ~ bottle)
  res <- tukey(obj)
  res

Estimate uncertainty from interlaboratory characterization

Description

Calculates the unweighted mean of laboratory means and its standard error. Common calibration or group-correlated components should be supplied separately to uncertainty_combine() so they are not counted repeatedly.

Usage

uncertainty_characterization(
  data,
  laboratory,
  result,
  name = "characterization",
  relative = FALSE
)

Arguments

Value

An uncertainty_characterization object.

References

CLSI EP30-A; CLSI EP32-R.

Examples

characterization_data <- data.frame(
  laboratory = rep(paste0("Lab", 1:4), each = 3),
  result = c(99, 100, 101, 100, 101, 100,
             98, 99, 100, 101, 102, 101)
)
characterization <- uncertainty_characterization(
  characterization_data, "laboratory", "result"
)
characterization$laboratory_summary

Combine standard uncertainty components

Description

Applies the GUM first-order covariance propagation formula ⁠c' Sigma c⁠. Components from any estimator in this module can be supplied directly.

Usage

uncertainty_combine(
  ...,
  sensitivity = NULL,
  correlation = NULL,
  value = NULL,
  coverage = 0.95,
  k = NULL
)

Arguments

Value

An uncertainty_budget object.

References

CLSI EP29-A.

Examples

repeatability <- uncertainty_type_a(
  c(9.9, 10.0, 10.1, 10.0), "repeatability", quantity = "mean"
)
calibrator <- uncertainty_type_b(
  "calibrator", estimate = 10, uncertainty = 0.2, k = 2,
  distribution = "normal"
)
budget <- uncertainty_combine(repeatability, calibrator, value = 10)
budget$components

Estimate the between-unit homogeneity uncertainty of a reference material

Description

Uses the EP30 one-way ANOVA calculations, including the effective replicate count for an unbalanced design and the inhomogeneity that can be hidden by measurement repeatability. The reported component is the larger of the observed between-unit SD and the hidden-inhomogeneity estimate.

Usage

uncertainty_homogeneity(
  data,
  unit,
  result,
  name = "between-unit homogeneity",
  relative = FALSE
)

Arguments

Value

An uncertainty_homogeneity object.

References

CLSI EP30-A.

Examples

homogeneity_data <- data.frame(
  unit = rep(paste0("V", 1:4), each = 2),
  result = c(99, 100, 101, 100, 98, 99, 102, 101)
)
homogeneity <- uncertainty_homogeneity(
  homogeneity_data, "unit", "result"
)
homogeneity$component

Estimate top-down uncertainty from internal quality-control data

Description

Uses a one-way random-effects ANOVA when runs contain replicates. The object reports within-run, between-run, single-result, and overall-mean estimates.

Usage

uncertainty_iqc(
  data,
  result,
  run = NULL,
  quantity = c("single", "mean"),
  name = "IQC",
  relative = FALSE
)

Arguments

Value

An uncertainty_iqc object.

References

CLSI EP29-A.

Examples

iqc_data <- data.frame(
  run = rep(1:5, each = 2),
  result = c(99, 101, 100, 102, 98, 100, 101, 103, 99, 100)
)
iqc <- uncertainty_iqc(iqc_data, "result", run = "run")
iqc$variance_components

Propagate uncertainty through a general measurement model

Description

The delta method numerically estimates sensitivity coefficients and applies GUM first-order propagation. Monte Carlo propagation samples the stored component distributions and returns a central quantile interval.

Usage

uncertainty_propagate(
  model,
  components,
  method = c("delta", "monte_carlo"),
  correlation = NULL,
  simulations = 100000L,
  coverage = 0.95,
  seed = NULL
)

Arguments

Value

An uncertainty_propagation object.

References

CLSI EP29-A.

Examples

numerator <- uncertainty_type_b(
  "numerator", estimate = 20, uncertainty = 0.4,
  distribution = "standard"
)
denominator <- uncertainty_type_b(
  "denominator", estimate = 10, uncertainty = 0.1,
  distribution = "standard"
)
propagated <- uncertainty_propagate(
  function(numerator, denominator) numerator / denominator,
  list(numerator, denominator), method = "delta"
)
propagated$estimate

Estimate long-term stability uncertainty

Description

Fits result on time and calculates the standard uncertainty at the requested shelf life as the slope standard error multiplied by shelf life. Separate components are returned for each optional storage condition.

Usage

uncertainty_stability(
  data,
  time,
  result,
  shelf_life,
  condition = NULL,
  name = "long-term stability",
  relative = FALSE
)

Arguments

Value

An uncertainty_stability object.

References

CLSI EP30-A.

Examples

stability_data <- data.frame(
  time = rep(0:4, each = 2),
  result = c(100.1, 99.9, 100.0, 99.8, 99.7, 99.9,
             99.6, 99.8, 99.5, 99.7)
)
stability_u <- uncertainty_stability(
  stability_data, "time", "result", shelf_life = 6
)
stability_u$summary

Accumulate uncertainty along a metrological traceability chain

Description

Accumulate uncertainty along a metrological traceability chain

Usage

uncertainty_traceability(stages, value = NULL, coverage = 0.95, k = NULL)

Arguments

Value

An uncertainty_traceability object.

References

CLSI EP29-A; CLSI EP32-R.

Examples

calibration <- uncertainty_type_b(
  "calibration", estimate = 100, uncertainty = 1, k = 2,
  distribution = "normal"
)
transport <- uncertainty_type_b(
  "transport", estimate = 100, lower = 99.5, upper = 100.5,
  distribution = "rectangular"
)
chain <- uncertainty_traceability(
  list(calibration = list(calibration), transport = list(transport)),
  value = 100
)
chain$summary

Estimate a Type A uncertainty component

Description

Uses the sample standard deviation for a future single result and the standard error for a mean. No normality or fitness-for-purpose decision is made.

Usage

uncertainty_type_a(x, name, quantity = c("single", "mean"), relative = FALSE)

Arguments