calculate_pod_quantiles: Point of Departure Estimation

In k-t-to/gravee: Good Risk Assessment Values for Environmental Exposures

Description

Estimate a chemical Point of Departure (POD) from dose-response data using bootstrap resampling and spline meta-regression

Usage

calculate_pod_quantiles(
  dose_response_data,
  resample_size = 1000,
  interpolation_size = 50,
  quantile_probs = c(0.025, 0.5, 0.975)
)

Arguments

dose_response_data	Data frame with two numeric columns. The first column should contain doses and the second column should contain the responses.
resample_size	The number of bootstrap samples
interpolation_size	The number of doses to interpolate between the minimum and maximum doses for spline prediction
quantile_probs	The quantiles to report for POD predictions

Details

Analysis is performed on log10-transformed doses. The returned PODs are back transformed to the original dose scale.

If the input data contains a dose = 0, the 0 dose is converted to be 1/10th the minimum non-zero dose, such that after log10-transformation, the distance between the 0 dose and minimum non-zero dose is 1. For PODs and interpolated doses less than the minimum non-zero dose, values are back-transformed and rescaled to be left bound by 0. Thus, in the output, for interpolated doses less than the minimum non-zero dose, values in the log10_dose column will not be equal to log10 of the dose column.

POD estimation assumes that the input dose-response data are asymptotic at low-doses. If the lower limit of the POD credible interval is within the two smallest doses, the function will return a warning: POD Credible Interval violates low-dose asymptote assumption. In this case, verify that an asymptote has been established at lower doses of the input data.

Value

The returned object prints the POD quantiles to the console. The returned object is a gravee list with four named components. The first component pods is a vector of length resample_size containing the estimated PODs from all bootstrap samples, on the original dose scale. The second element pod_quantiles contains the POD quantiles, with boundaries defined by quantile_probs. The third component all_res is a list of length resample_size containing the interpolated spline predictions from each bootstrap sample and curvature measurements along the fit curve. The fourth component bs_samples is a list of length resample_size containing the sampled values for each bootstrap sample.

Examples

x <- rep(c(0,0.05,0.1,0.5,1), each = 3)
y <- 1/exp(-x) + abs(rnorm(length(x)))
df <- data.frame(x,y)
pods <- calculate_pod_quantiles(df)
plot(pods)

k-t-to/gravee documentation built on March 26, 2021, 3:10 a.m.