DR.fit: Fit a Dose/Response function

In robertogattabs/RadAgent: Package for handling DICOM RT files

Description

This function fits an object of class dvhatrix or numeric to a given series of outcome.

Usage

DR.fit(doses, outcome, DR.fun = c("Lyman", "Niemierko", "Bentzen", "Goitein",
  "Munro", "Okunieff", "Warkentin"), type = c("NTCP", "TCP"), CI = TRUE,
  C.I.width = 0.95, epsilon = 1e-06)

Arguments

doses	Either a dvhmatrix class object or a vector with nominal doses
outcome	A numeric vector of cases showing (1) and not showing (0) the outcome
DR.fun	Dose/Response function, a character vector containing the name of one of the function in the package moddicom: Lyman, Niemierko, Bentzen, Goitein, Munro, Okunieff, Warkentin.
type	Function type: NTCP, Normal Tissue Complication Probability, or TCP, Tumor Control Probability
CI	If TRUE it returns the value of confidence interval calulated by profile likelihood method
epsilon	Lower interval for bisecting the MLE
CI.width	The value of width of confidence interval to be returned if CI = TRUE

Details

This function is a wrapper for fitting dose/response functions to a given set of binary outcomes. Currently outcomes must be referred as a numeric vector representing cases "showing the outcome" with the number 1, or "not showing the outcome" with the number 0. Model fitting is performed using Maximum Likelihood Estimation method, implemented by wrapping R MLE functions set for specific Log Likelihood functions related to dose/response formulas. If doses is a vector of nominal doses its length must be equal to the length of outcome, if doses is a dvhmatrix class object the number of histograms must be equal to the length of outcome.

robertogattabs/RadAgent documentation built on June 30, 2018, 12:02 a.m.