NEWS

---

diagmeta 0.6.0

Major changes

• Data sets with negative association between biomarker values and probability of target condition can be used without data manipulation

• Order of panels in plot.diagmeta() can be specified by the user

Bug fixes

• plot.diagmeta():
  • use chull() to determine the correct order of x- and y-values to print confidence regions for sensitivity and specificity in the SROC curve

User-visible changes

• diagmeta(), ipd2diag();

  • new argument 'direction' to specify whether the probability of the target condition (e.g., a disease) is increasing or decreasing with higher values of the biomarker

• ipd2diag();

  • new argument 'data' to provide data set with information on arguments 'studlab', 'value', and 'status'

• diagmeta():

  • input to first argument 'TP' can be an object created with ipd2diag()

• plot.diagmeta():

  • default set of graphs (argument 'which') changed from c("survival", "youden", "roc", "sroc") to c("regression", "cdf", "sensspec","youden", "roc", "sroc")
  • new argument 'ylim' to specify the y-limits of all plot (selected plots must have a similar y-axis)

Internal changes

• New internal function plot.diagmeta-internal()

• ipd2diag();

  • additional class 'ipd2diag' and attribute 'direction'

diagmeta 0.5-1 (2022-12-21)

User-visible changes

• Change maintainer's email address

• New branch 'release' on GitHub starting with diagmeta, version 0.5-1

Internal changes

• Rename list element 'Cov.fixed' to 'Cov.common'

diagmeta 0.5-0 (2022-04-22)

Major changes

• Behaviour of print.diagmeta() and print.summary.diagmeta() switched (to be in line with other print and summary functions in R)

• Do not stop with an error if optimal cut-off cannot be determined for logistic distribution

• Calculate area under the curve for specificity given sensitivity

Bug fixes

• diagmeta():
  • fix for erratic confidence limits of AUC which could be outside the admissible range from 0 to 1 or exclude the AUC estimate

User-visible changes

• More concise printout for summary.diagmeta()

Internal changes

• diagmeta():

  • new list elements 'AUCSens' and 'AUCSpec' to calculate AUC for sensitivity given specificity or vice versa (existing list element 'AUC' is equal to 'AUCSens')

• New internal function catch() to catch value for an argument

diagmeta 0.4-1 (2021-05-11)

Bug fixes

• plot.diagmeta():

  • print correct confidence region for specificities in SROC curves

• diagstats():

  • print results for requested specificity if only argument 'spec' is provided

User-visible changes

• Use Markdown for NEWS

Internal changes

• diagmeta():
  • new list element 'Cov.fixed' with covariance matrix from fixed effects model

diagmeta 0.4-0 (2020-04-02)

Major changes

• New default model (argument 'model') in diagmeta(), i.e., common random intercept and common slope ("CICS"), due to estimation problems with the previous default ("DIDS") after changes in R package lme4

Bug fixes

• plot.diagmeta():
  • correct line types for survival functions

User-visible changes

• diagmeta():

  • print a more informative error message in case of a negative correlation between increasing cutoffs and sensitivity

• plot.diagmeta():

  • argument 'points' considered for plots of type "regression", "cdf", "survival", "Youden", "ROC" and "sensspec"

• Help pages:

  • use the default model in all examples

diagmeta 0.3-1 (2019-04-11)

User-visible changes

• Export R functions:

  • as.data.frame.diagmeta(), plot.diagmeta(), print.diagmeta(), print.diagstats(), print.summary.diagmeta(), summary.diagmeta()

• plot.diagmeta():

  • argument 'col.points' can be any color defined in colours()
  • new argument 'col.ci' to specify color of curves with confidence limits

Internal changes

• diagmeta():
  • check for numerical values in arguments 'TP', 'FP', 'TN', 'FN', and 'cutoff'

diagmeta 0.3-0 (2018-12-11)

User-visible changes

• plot.diagmeta(): new plot type to show sensitivity and specificity curves

• New arguments 'sens' and 'spec' in diagstats()

• print.summary.diagmeta():

  • print confidence interval for optimal cutoff (for normal distribution)

• New function as.data.frame.diagmeta()

Bug fixes

• plot.diagmeta():
  • correct ROC curves for datasets with decreasing cutoff values for individual studies (points (0, 0) and (1, 1) were connected with the wrong values on the ROC curve)

Internal changes

• diagmeta():
  • calculate and return lower and upper confidence limit for optimal cutoff (for normal distribution)

diagmeta 0.2-0 (2018-03-23)

First version released on CRAN