-
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
- 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
-
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)
-
New internal function plot.diagmeta-internal()
-
ipd2diag();
- additional class 'ipd2diag' and attribute 'direction'
-
Change maintainer's email address
-
New branch 'release' on GitHub starting with diagmeta, version 0.5-1
- Rename list element 'Cov.fixed' to 'Cov.common'
-
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
- diagmeta():
- fix for erratic confidence limits of AUC which could be outside the admissible range from 0 to 1 or exclude the AUC estimate
- More concise printout for summary.diagmeta()
-
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
-
plot.diagmeta():
- print correct confidence region for specificities in SROC curves
-
diagstats():
- print results for requested specificity if only argument 'spec' is provided
- Use Markdown for NEWS
- diagmeta():
- new list element 'Cov.fixed' with covariance matrix from fixed effects model
- 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
- plot.diagmeta():
- correct line types for survival functions
-
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
-
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
- diagmeta():
- check for numerical values in arguments 'TP', 'FP', 'TN', 'FN', and 'cutoff'
-
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()
- 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)
- diagmeta():
- calculate and return lower and upper confidence limit for optimal cutoff (for normal distribution)