The function fits the model of Holling et al. (2012). The adjusted profile maximum likelihood estimator (APMLE) is implemented for homogeneity and heterogeneity of primary studies.
phm(data, ...)
# S3 method for default
phm(data = NULL, subset=NULL,
TP="TP", FN="FN", FP="FP", TN="TN",
correction = 0.5, correction.control = "all",
hetero = TRUE, estimator = "APMLE", l = 100, ...)any object that can be converted to a data frame with integer variables TP, FN, FP and TN, alternatively a matrix with column names including TP, FN, FP and TN.
the rows of data to be used as a subset in all calculations. If NULL (the default) then the complete data is considered.
character or integer: name for vector of integers that is a variable of data or a vector of integers. If data is not NULL, names are expected, otherwise integers are.
character or integer: name for vector of integers that is a variable of data or a vector of integers. If data is not NULL, names are expected, otherwise integers are.
character or integer: name for vector of integers that is a variable of data or a vector of integers. If data is not NULL, names are expected, otherwise integers are.
character or integer: name for vector of integers that is a variable of data or a vector of integers. If data is not NULL, names are expected, otherwise integers are.
numeric, continuity correction applied if zero cells
character, if set to "all" (the default) the continuity correction is added to the whole data if only one cell in one study is zero. If set to "single" the correction is only applied to rows of the data which have a zero.
logical, should heterogeneity of studies be assumed? Will fit model for homogeneity otherwise.
character, determines estimator used. Currently only APMLE is available.
interger, number of iterations for fixed point algorithm
arguments passed on to other functions (currently not used)
An object of the class phm for which many standard methods are available. See phm-class for details.
The model of Holling et al. (2012) assumes that the relationship between false positive rates \(u\) and and sensitivities \(p\) can be described by $$u^\theta = p,$$ where \(\theta\) is the diagnostic accuracy parameter. If homogeneity of the studies can be assumed, \(\theta\) is estimated as a fixed effect. Under heterogeneity a random effect with variance \(\tau^2\) describes the variation of the diagnostic accuracy parameter in the population of studies. Since the error of each observed \(\theta\) depends only on the sample size and \(\theta\) the model has only one parameter in the case of homogeneity and two parameters under heterogeneity, making it suitable for diagnostic meta-analysis with low sample size. Estimation proceeds by a fixed point algorithm derived from the adjusted profile likelihood. More details on the computational approach can be found in Holling et al. (2012).
Holling, H., Boehning W., Boehning, D. (2012) “Meta-Analysis of Diagnostic Studies based upon SROC-Curves: a Mixed Model Approach using a Proportional Hazards Model.” Statistical Modelling, 12, 347???-375.
# NOT RUN {
data(AuditC)
(fit <- phm(AuditC))
summary(fit)
plot(fit)
# }
Run the code above in your browser using DataLab