Skip to contents

Linearity Assessment

Source: R/accuracy.R
measure_linearity.Rd

Assesses linearity of a method by evaluating the relationship between response and concentration across the specified range.

Usage

measure_linearity(
  data,
  conc_col,
  response_col,
  method = c("regression", "residual"),
  conf_level = 0.95
)

Arguments

data

A data frame containing concentration and response data.

conc_col

Name of the column containing concentrations.

response_col

Name of the column containing responses.

method

Linearity assessment method:

  • "regression" (default): Linear regression with diagnostics

  • "residual": Residual analysis and lack-of-fit test

conf_level

Confidence level for intervals. Default is 0.95.

Value

A measure_linearity object containing:

  • r_squared: Coefficient of determination

  • adj_r_squared: Adjusted R-squared

  • slope: Regression slope with CI

  • intercept: Regression intercept with CI

  • residual_sd: Residual standard deviation

  • lack_of_fit: Lack-of-fit test results (if replicates exist)

  • range: Concentration range evaluated

Details

Linearity demonstrates that the method produces results that are directly proportional to analyte concentration within a given range.

Assessment Criteria

  • R-squared >= 0.99 (typical for many applications)

  • Residuals randomly distributed around zero

  • No systematic pattern in residual plots

  • Lack-of-fit test not significant (p > 0.05)

ICH Q2 Requirements

Linearity should be evaluated across the range with at least 5 concentration levels. Report the regression equation, correlation coefficient, and visual inspection of residual plots.

See also

measure_accuracy(), measure_calibration_fit()

Other accuracy: measure_accuracy(), measure_carryover()

Examples

# Linearity assessment
set.seed(123)
data <- data.frame(
  concentration = rep(c(10, 25, 50, 75, 100), each = 3),
  response = rep(c(10, 25, 50, 75, 100), each = 3) * 1.5 + rnorm(15, 0, 2)
)

result <- measure_linearity(data, "concentration", "response")
print(result)
#> measure_linearity
#> ──────────────────────────────────────────────────────────────────────────────── 
#> 
#> Data:
#>   n = 15 ( 5 levels )
#>   Range: 10 - 100 
#> 
#> Regression:
#>   Slope = 1.493 
#>     95% CI: [1.463, 1.523]
#>   Intercept = 0.675 
#>     95% CI: [-1.147, 2.497]
#> 
#> Fit Quality:
#>   R-squared = 0.9989 
#>   Adj. R-squared = 0.99882 
#>   Residual SD = 1.737 
#>   Residual CV = 2.2 %
#> 
#> Lack-of-Fit Test:
#>   F = 0.877 
#>   p-value = 0.4853 
#>   Result: Not significant (linearity acceptable)