Package 'referenceIntervals'

Title: Reference Intervals
Description: This is a collection of tools to allow the medical professional to calculate appropriate reference ranges (intervals) with confidence intervals around the limits for diagnostic purposes.
Authors: Daniel Finnegan
Maintainer: Daniel Finnegan <[email protected]>
License: GPL-3
Version: 1.3.1
Built: 2026-06-03 09:17:15 UTC
Source: https://github.com/cran/referenceIntervals

Help Index

This package calculates reference intervals from a dataset using either parametric, non-parametric, or robust methods.

Description

This package also calculates the confidence intervals around the calculated reference intervals in order to provide a metric for how precise the calculations are. This package also contains four outlier detection functions.

Details

Package: referenceIntervals
Type: Package
Version: 1.3.1
Date: 2024-03-30
License: GPL-3

Author(s)

Daniel Finnegan

Maintainer: Daniel Finnegan <[email protected]>

References

Clinical and Laboratory Standards Institute. Defining, Establishing, and Verifying Reference Intervals in the Clinical Laboratory; Approved Guideline - Third Edition. C28-A3c. 28(30).

Friedman JH. A recursive partitioning decision rule for nonparametric classification. IEEE Transactions on Computers. 1977. April: 404-408.

Friedrichs KR, Harr KE, Freeman KP, et al. ASVCP reference interval guidelines: determination of de novo reference intervals in veterinary species and other related topics. Vet Clin Pathol. 2012;441-453.

H.E. Solberg Department of Clinical Chemistry. Approved recommendation (1987) on the theory of reference values. Part 5. Statistical treatment of collected reference values. Determination of reference limits. Clinica Chimica Acta. 1987:S13-S32.

Horn PS, Pesce AJ. Reference intervals: an update. Clinica Chimica Acta. 2003. 334:5-23.

Horn PS, Pesce AJ, Copeland BE. A robust approach to reference interval estimation and evaluation. Clinical Chemistry. 1998. 44(3):622-631.

Horn PS, Pesce AJ, Copeland BE. Reference interval computation using robust vs parametric and nonparametric analyses. Clinical Chemistry. 1999. 45(12):2284-2285.

Linnet K. Nonparametric estimation of reference intervals by simple and bootstrap-based procedures. Clinical Chemistry. 2000. 46(6):867-869.

Verma SP, Quiroz-Ruiz A. Critical values for six Dixon tests for outliers in normal samples up to sizes 100, and applications in science and engineering. Revista Mexicana de Ciencias Geologicas. 2006. 23(2):133-161.

Verma SP, Quiroz-Ruiz A, Diaz-Gonzalez L. Critical values for 33 discordancy test variants for outliers in normal samples up to sizes 1000, and applications in quality control in Earth Sciences. Revista Mexicana de Ciencias Geologicas. 2008. 25(1):82-96.

Virtanen A, Kairisto V, Uusipaikka E. Regression-based reference limits: determination of sufficient sample size. Clinical Chemistry. 1998. 44(11):2353-2358.

Examples

refLimit(set50, out.rm = TRUE, out.method = "cook")
refLimit(set200, out.method = "horn", RI = "n", refConf = 0.90, limitConf = 0.80)
horn.outliers(set120)
dixon.outliers(set20)

refLimit(set50, out.method = "vanderLoo", out.rm = TRUE, RI = "r", bootStat = "perc")

frame = data.frame(one = rnorm(30, m = 5, sd = 2), two = rnorm(30, m = 7, sd = 1),
	three = rnorm(30, m = 2, sd = 0.5))
result = refLimit(frame)
plot(result)

Determines outliers using Cook's Distance

Description

A linear regression model is calculated for the data (which is the mean for one-dimensional data. From that, using the Cook Distances of each data point, outliers are determined and returned.

Usage

cook.outliers(data)

Arguments

data

A vector of data points.

Value

Returns a list containing a vector of outliers and a vector of the cleaned data (subset).

outliers

A vector of outliers from the data set

subset

A vector containing the remaining data, cleaned of outliers

Author(s)

Daniel Finnegan

Examples

cook.outliers(set50)
plot(cook.outliers(set50)$subset)

Determines outliers using Dixon's Q Test method

Description

This determines outliers of the dataset by calculating Dixon's Q statistic and comparing it to a standardized table of statistics. This method can only determine outliers for datasets of size 3 <= n <= 30. This function requires the outliers package.

Usage

dixon.outliers(data)

Arguments

data

A vector of data points.

Value

Returns a list containing a vector of outliers and a vector of the cleaned data (subset).

outliers

A vector of outliers from the data set

subset

A vector containing the remaining data, cleaned of outliers

Author(s)

Daniel Finnegan

References

Statistical treatment for rejection of deviant values: critical values of Dixon's "Q" parameter and related subrange ratios at the 95 (2), pp 139-146 DOI: 10.1021/ac00002a010. Publication Date: January 1991

One-sided and Two-sided Critical Values for Dixon's Outlier Test for Sample Sizes up to n = 30. Economic Quality Control, Vol 23(2008), No. 1, 5-13.

Examples

dixon.outliers(set20)
summary(dixon.outliers(set20)$subset)

Determines outliers using Horn's method and Tukey's interquartile fences on a Box-Cox transformation of the data.

Description

This function determines outliers in a Box-Cox transformed dataset using Horn's method of outlier detection using Tukey's interquartile fences. If a data point lies outside 1.5 * IQR from the 1st or 3rd quartile point, it is an outlier.

Usage

horn.outliers(data)

Arguments

data

A vector of data points.

Value

Returns a list containing a vector of outliers and a vector of the cleaned data (subset).

outliers

A vector of outliers from the data set

subset

A vector containing the remaining data, cleaned of outliers

Author(s)

Daniel Finnegan

References

ASVCP reference interval guidelines: determination of de novo reference intervals in veterinary species and other related topics. Vet Clin Pathol 41/4 (2012) 441-453, 2012 American Society for Veterinary Clinical Pathology

Horn, P. S., Feng, L., Li, Y., & Pesce, A. J. (2001). Effect of outliers and nonhealthy individuals on reference interval estimation. Clinical Chemistry, 47(12), 2137-2145.

Horn, P.S., Pesce, A.J. (2003). Reference Intervals: an update. Clin Chim Acta. 334(1-2):5-23. DOI: doi: 10.1016/s0009-8981(03)00133-5.

Examples

horn.outliers(set200)

Table that dictate the ranks for the confidence intervals around the calculated reference interval.

Description

This is a table that dictate the ranks for the confidence intervals around the calculated reference interval. This method is available when 120 <= n <= 1000.

Usage

nonparRanks

Format

A data frame with 882 observations on the following 3 variables.

SampleSize

a numeric vector

Lower

a numeric vector

Upper

a numeric vector

References

Defining, Establishing, and Verifying Reference Intervals in the Clinical Laboratory; Approved Guideline - 3rd Edition (C28-A3)

Examples

data(nonparRanks)

Determines the reference interval using non-parametric means

Description

This function uses the appropriate percentiles as determined by refConf to return the non-parametric reference interval. This is written as a boot function to use within the function refLimit.

Usage

nonparRI(data, indices = 1:length(data), refConf = 0.95)

Arguments

data

data is a vector of sample values.

indices

The indices of data to be used in the calculations. The default is to use the whole set.

refConf

refConf is a measure of the range covered by the calculation. Most often, as is the default, 95

Value

Returns a vector of two values, the lower and upper limits of the reference interval.

Author(s)

Daniel Finnegan

References

Defining, Establishing, and Verifying Reference Intervals in the Clinical Laboratory; Approved Guideline - 3rd Edition (C28-A3)

Examples

nonparRI(set50)
nonparRI(set50, refConf = 0.9)

Overload of plot function to include the ability to plot the results of refLimit

Description

Plots the reference interval and confidence intervals around the limits of the reference interval.

Usage

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

Arguments

x

Object x is of type "interval".

main

Title of plot.

...

Arguments to be passed to methods, such as graphical parameters (see 'par').

Value

No return value.

Author(s)

Daniel Finnegan

Examples

result = refLimit(set200)
plot(result)

Overload of print in order to concisely print the results of refLimit

Description

This function allows for the pretty-printing of a large list object created by calling the refLimit function.

Usage

## S3 method for class 'interval'
print(x, digits = 4L, quote = TRUE, prefix = "", ...)

Arguments

x

x is an object of type "interval"

digits

minimal number of _significant_ digits. See 'print.default'.

quote

logical, indicating whether or not strings should be printed with surrounding quotes.

prefix

Option to specify a formatting prefix.

...

further arguments passed to or from other methods.

Value

No return value.

Author(s)

Daniel Finnegan

Examples

result = refLimit(set120)
result

Overload of print in order to concisely print the results of refLimit

Description

This function allows for the pretty-printing of a large list object created by calling the refLimit function.

Usage

## S3 method for class 'interval.sub'
print(x, digits = 4L, quote = TRUE, prefix = "", ...)

Arguments

x

x is an object of type "interval"

digits

minimal number of _significant_ digits. See 'print.default'.

quote

logical, indicating whether or not strings should be printed with surrounding quotes.

prefix

Option to specify a formatting prefix.

...

further arguments passed to or from other methods.

Value

No return value.

Author(s)

Daniel Finnegan

Calculates and returns reference and confidence intervals for a dataset

Description

This function calculates a reference interval from a dataset using parametric, non-parametric, or robust methods.

Usage

refLimit(data, out.method = "horn", out.rm = FALSE, RI = "p", CI = "p",
refConf = 0.95, limitConf = 0.9, bootStat = "basic")

Arguments

data

A vector of data points.

out.method

The outlier detection method. Valid options include "horn", "cook", "dixon", and "vanderLoo".

out.rm

Remove outliers. If value is TRUE, outliers will be automatically removed prior to calculations. If FALSE (default), outliers will be detected but not removed.

RI

Method for reference interval calculations. Valid options include "p" (default) for parametric, "n" for non-parametric, and "r" for robust method.

CI

Method for confidence interval calculations. Valid options include "p" for parametric (default), "n" for non-parametric, and "boot" for bootstrapping method. The minimum samplesize for non-parametric confidence interval calculations is 120. With smaller samples, bootstrapping methods are used.

refConf

Desired coverage for the calculated reference interval. The default is a 95 interval.

limitConf

Desired confidence interval level. The default is a 90 reference interval limits.

bootStat

Method for calculating confidence intervals from package boot. Valid arguments include "basic" (basic bootstrap method), "perc" (bootstrap percentile method), "norm" (normal approximation method), "stud" (studentized bootstrap method), and "bca" (adjusted bootstrap percentile method).

Details

A confidence interval around each limit of the reference interval is calculated as a metric for determining the validity of the result. Outliers can be detected in one of four different methods and automatically eliminated.

To determine the most appropriate calculation for confidence intervals using the bootstrapping method, please consult chapter 5 of Davison and Hinkley's "Bootstrap Methods and their Applications."

Value

Returns a list of necessary information.

size

Size of dataset

dname

Name of dataset

out.method

Outlier detection method

out.rm

Boolean indicating whether outliers are automatically removed

outliers

Vector of detected outliers

methodRI

Method for reference interval calculations (p, n, or r)

methodCI

Method for confidence interval calculations (p, n, boot)

norm

Results of running Shapiro-Wilk and Kolmorgorov-Smirnov normacy tests

refConf

Desired coverage of reference interval

limitConf

Desired confidence interval level

Ref_Int

List containing the reference interval and confidence interval values

Author(s)

Daniel Finnegan

References

ASVCP reference interval guidelines: determination of de novo reference intervals in veterinary species and other related topics. Vet Clin Pathol 41/4 (2012) 441-453, 2012. American Society for Veterinary Clinical Pathology

Davison, A.C. and Hinkley, D.V. (1997) Bootstrap Methods and Their Application, Chapter 5. Cambridge University Press.

Examples

refLimit(set20, out.method = "dixon")
refLimit(set200, out.method = "cook", out.rm = TRUE, RI = "n", refConf = 0.9)
refLimit(set50, out.method = "vanderLoo", out.rm = TRUE, RI = "r", bootStat = "perc")

Algorithm that implements the robust method for reference interval calculations

Description

The robust method is an iterative method that determines the most appropriate weighted mean of the data and then calculates the desired reference interval.

Usage

robust(data, indices = c(1:length(data)), refConf = 0.95)

Arguments

data

Vector of data.

indices

Indices of data to use for calculations.

refConf

Desired coverage of the reference interval. Default is 95 interval.

Value

Returns a vector containing the lower and upper limits of the reference interval.

Author(s)

Daniel Finnegan

References

Defining, Establishing, and Verifying Reference Intervals in the Clinical Laboratory; Approved Guideline - 3rd Edition (C28-A3)

Examples

robust(set50)
robust(horn.outliers(set20)$subset)

Dataset containing 120 values

Description

Small dataset containing 120 samples. The mean is centered on 27 with a standard deviation of 7.

Usage

set120

Format

The format is: num [1:120] 38.1 12.6 31.3 35.5 22.6 ...

Source

rnorm(120, m = 27, sd = 7)

Examples

data(set120)

Small dataset containing 20 samples

Description

Small dataset containing 20 samples. The mean is centered on 42 with a standard deviation of 5.

Usage

set20

Format

The format is: num [1:20] 35 32.9 43.6 44.6 35.9 ...

Source

rnorm(20, m = 42, sd = 6)

Examples

data(set20)

Dataset containing 200 values

Description

Small dataset containing 200 samples. The mean is centered on 5 with a standard deviation of 1.

Usage

set200

Format

The format is: num [1:200] 3.95 5.16 5.32 3.86 3.54 ...

Source

rnorm(200, m = 5, sd = 1)

Examples

data(set200)

Dataset containing 50 values

Description

Small dataset containing 50 samples. The mean is centered on 14 with a standard deviation of 3.

Usage

set50

Format

The format is: num [1:50] 16.61 20.43 7.91 15.19 14.77 ...

Source

rnorm(50, m = 14, sd = 3)

Examples

data(set50)

This is the workhorse of the refLimit function

Description

This is the function called to work on each individual vector of data.

Usage

singleRefLimit(data, dname = "default", out.method = "horn", out.rm = FALSE,
RI = "p", CI = "p", refConf = 0.95, limitConf = 0.9, bootStat = "basic")

Arguments

data

A vector of data points.

dname

Name of dataset.

out.method

The outlier detection method. Valid options include "horn", "cook", "dixon", and "vanderLoo".

out.rm

Remove outliers. If value is TRUE, outliers will be automatically removed prior to calculations. If FALSE (default), outliers will be detected but not removed.

RI

Method for reference interval calculations. Valid options include "p" (default) for parametric, "n" for non-parametric, and "r" for robust method.

CI

Method for confidence interval calculations. Valid options include "p" for parametric (default), "n" for non-parametric, and "boot" for bootstrapping method. The minimum sample size for non-parametric confidence interval calculations is 120. With smaller samples, bootstrapping methods are used.

refConf

Desired coverage for the calculated reference interval. The default is a 95

limitConf

Desired confidence interval level. The default is a 90 confidence interval around the reference interval limits.

bootStat

Method for calculating confidence intervals from package boot. Valid arguments include "basic" (basic bootstrap method), "perc" (bootstrap percentile method), "norm" (normal approximation method), "stud" (studentized bootstrap method), and "bca" (adjusted bootstrap percentile method).

Value

Returns a list of necessary information.

size

Size of dataset

dname

Name of dataset

out.method

Method of outlier detection

out.rm

Boolean indicating whether outliers are automatically removed

outliers

Vector of detected outliers

methodRI

Method for reference interval calculations (p, n, or r)

methodCI

Method for confidence interval calculations (p, n, boot)

norm

Results of running Shapiro-Wilk and Kolmorgorov-Smirnov normacy tests

refConf

Desired coverage of reference interval

limitConf

Desired confidence interval level

Ref_Int

List containing the reference interval and confidence interval values

Author(s)

Daniel Finnegan

Examples

singleRefLimit(set200, out.method = "horn", out.rm = TRUE)

Mark van der Loo's outlier detection method in the extremevalues package

Description

Separates data into vectors of outliers and a cleaned subset of the data.

Usage

vanderLoo.outliers(data)

Arguments

data

Vector of data values.

Value

Returns a list containing a vector of outliers and a vector of the cleaned data (subset).

outliers

A vector of outliers from the data set

subset

A vector containing the remaining data, cleaned of outliers

Note

Requires extremevalues package.

Author(s)

Daniel Finnegan

References

http://cran.r-project.org/web/packages/extremevalues/extremevalues.pdf

Examples

vanderLoo.outliers(set50)
vanderLoo.outliers(set200)