correct.trees: Correct the Classification Outcomes by Methods of Botella et...

In Metatron: Meta-analysis for Classification Data and Correction to Imperfect Reference

Description

This function corrects the frequencies reported in the primary studies when a imperfect reference was used in the classification process. According to the conjoint parameter estimates of multinomial tree model 2 by Botella et al (2013) given by the imperfect.trees, one can obtain the corrected frequencies as if the a perfect standard reference were used. Note that correction should be used only if the model 2 is the optimal one compared to model 1.

Usage

correct.trees(x,TP,FN,TN,FP,study,data)

Arguments

x	an object of class "imperfect.trees".
TP	the true positive counts reported in primary studies.Vector of integers,need to be speficied either directly or by referring to a variable in data frame.
FN	the false negative counts reported in primary studies.Vector of integers,need to be speficied either directly or by referring to a variable in data frame.
TN	the true negative counts reported in primary studies.Vector of integers,need to be speficied either directly or by referring to a variable in data frame.
FP	the false positive counts reported in primary studies.Vector of integers,need to be speficied either directly or by referring to a variable in data frame.
study	study names or identities. Character vector, need to be speficied either directly or by referring to a variable in data frame.
data	optional data frame that contains the above-mentioned variables.

Details

The study names and their order used by this function to correct frequencies need to be exactly the same as in the previous step which fits the multinomial tree model 2 assuming an imperfect reference. Otherwise it will give an error indicating the problem.

Value

A data frame with the 4 corrected frequencies for each primary study, namely, TPnew, FNnew, FPnew, TNnew. If the input data contains other information such as covariates, these will be kept unchaged and well matched with the newly corrected frequencies. This data frame can be further used to fit the bivariate model of Reitsma et al. (2005) by the fit.bivar function.

Author(s)

Huiling Huang <[e-mail]>

References

Botella, J., Huang, H., Suero, M.(2013). Multinomial tree models for assessing the status of the reference in studies of the accuracy of tools for binary classification. Frontiers in Psychology.4:694. http://www.frontiersin.org/Journal/Abstract.aspx?s=956&name=quantitative_psychology_and_measurement&ART_DOI=10.3389/fpsyg.2013.00694

Botella, J., Huang, H., Suero, M. Meta-analysis of the accuracy of tools used for binary classification when the primary studies employ different references. To appear in Psychological Methods.

See Also

imperfect.trees.

Examples

## data of the screening tool Mini Mental State Examination (MMSE) from Botella et al.(2013)
data(MMSE)
## fit the multinomial tree model 2, imperfect reference
(mmse2<-imperfect.trees(TP=TP,FN=FN,TN=TN,FP=FP,study=study,data=MMSE))
## after comparing to the results of model 1, the model 2 is chosen, 
## then comes frequency correction. 
correct.trees(mmse2,TP=TP,FN=FN,TN=TN,FP=FP,study=study,data=MMSE)

Example output

Loading required package: lme4
Loading required package: Matrix
Loading required package: mpt
There are 4 independent trees and 8 parameters in this model. 

Estimation of the accuracy indices of both reference and the test of interest:
           Se_R       Se_T  Sp_R       Sp_T
[1,]  0.8759778  0.8640778     1  0.8716447

Estimation of prevalence in each primary study:
  Prevalence_1 Prevalence_2 Prevalence_3 Prevalence_4
1    0.1238975    0.3474615    0.5642283   0.04489575

Model fit statistics:
AIC= 88.49812 

         G2          df        pval 
12.13630310  4.00000000  0.01636595 

                 study  TP FN  FP  TN     TPnew     FNnew     FPnew    TNnew
1 Brayne&Calloway,1989  24  5  31 205  27.27369  5.559136  27.72631 204.4409
2   Brodaty et al,2002  66 16  48 153  80.77209 17.559518  33.22791 151.4405
3    Clarke et al,1991 137 17  28 122 151.54507 19.980330  13.45493 119.0197
4    Cullen et al,2005  40  4 138 933  45.21139  4.847382 132.78861 932.1526

Metatron documentation built on May 1, 2019, 10:20 p.m.