baclava (version 1.0)

predictODX: Estimate the Overall and Per Screen Overdiagnosis Rates

Description

Using the posterior parameter distributions, calculates the infinite population estimates of the probability of overdiagnosis at each screening episode due to indolence and/or death by other causes.

Usage

predictODX(
  object,
  screening.schedule,
  other.cause.rates,
  groups.rateP = NULL,
  screen.type = NULL,
  burnin = 1000L,
  verbose = TRUE
)
# S3 method for baclava.ODX.pred
plot(x, y, ...)

Value

A list object. For each screen in screening.schedule, a matrix providing the mean total overdiagnosis and the mean overdiagnosis due to indolent/progressive tumors, as well as their 95

Similarly, element overall provides these estimates for the full screening schedule.

Arguments

object

An object of S3 class 'baclava'. The value object returned by fit_baclava().

screening.schedule

A numeric vector object. A vector of ages at which screenings occur.

other.cause.rates

A data.frame object. Must contain columns "Rate" and "Age".

groups.rateP

An integer scalar object. If model included groups with different sojourn parameters, the group for which overdiagnosis is to be estimated. Must be one of object$setup$groups.rateP

screen.type

An integer scalar object. If model included screen-type, specific sensitivity parameters, the screen-type for which overdiagnosis is to be estimated. Must be one of object$setup$groups.beta

burnin

An integer object. Optional. The number of burn-in samples. Used only for type = "trace". One trace plot is generated for the burnin iterations; a second for the post-burnin iterations. Note, this refers to the kept (thinned) samples.

verbose

A logical object. If TRUE, progress bars will be displayed.

x

A an object of S3 class 'baclava.PDX.pred' as returned by predictODX().

y

Ignored.

...

Ignored.

Functions

• plot(baclava.ODX.pred): Generate column plot of predicted overdiagnosis for each screen.

Details

Provided birth cohort life table is an all cause tables obtained from the CDC Life Tables. Vital Statistics of the United States, 1974 Life Tables, Vol. II, Section 5. 1976. Estimated "other cause" mortality will thus be overestimated when using these tables. It is recommended that user provide data that has been corrected to exclude death due to the disease under analysis.

Examples

data(screen_data)

theta_0 <- list("rate_H" = 7e-4, "shape_H" = 2.0,
                "rate_P" = 0.5  , "shape_P" = 1.0,
                "beta" = 0.9, psi = 0.4)
prior <- list("rate_H" = 0.01, "shape_H" = 1,
              "rate_P" = 0.01, "shape_P" = 1,
              "a_psi" = 1/2 , "b_psi" = 1/2,
              "a_beta" = 38.5, "b_beta" = 5.8)

# This is for illustration only -- the number of MCMC samples should be
# significantly larger and the epsilon values should be tuned.
example <- fit_baclava(data.assess = data.screen,
                       data.clinical = data.clinical,
                       t0 = 30.0,
                       theta_0 = theta_0,
                       prior = prior)
                       
# if rates are not available, an all cause dataset is provided in the package
# NOTE: these predictions will be over-estimated
           
data(all_cause_rates)
all_cause_rates <- all_cause_rates[, c("Age", "both")]
colnames(all_cause_rates) <- c("Age", "Rate")

# using single screen for example speed
predicted_odx <- predictODX(object = example, 
                            other.cause.rates = all_cause_rates,
                            screening.schedule = 40, 
                            burnin = 10)

plot(predicted_odx)