Package 'radir'

Inverse regression model for radiation biodosimetry

Description

The package implements a new inverse regression model with applications to radiation biodosimetry.

Details

Package:	radir
Type:	Package
Version:	1.0.4
Date:	2019-6-30
License:	GPL version 2 or newer
LazyLoad:	yes

The package implements a new inverse regression model with applications to radiation biodosimetry by means of the function dose.distr. It allows for several distributions for the dose prior including uniform and gamma, and for the mean prior, including gamma and normal distributions. A summary containing the most relevant information about the estimated doses can be obtained via summary and graphics can be obtained in a standard way by means of plot or lines functions.

Author(s)

David Moriña (Barcelona Graduate School of Mathematics), Manuel Higueras (Basque Center for Applied Mathematics) and Pedro Puig (Universitat Autònoma de Barcelona)

Mantainer: David Moriña Soler <[email protected]>

References

Higueras M, Puig P, Ainsbury EA, Rothkamm K. A new inverse regression model applied to radiation biodosimetry. Proc R Soc A 2015;471, http://dx.doi.org/10.1098/rspa.2014.0588

See Also

dose.distr

Examples

f <- expression(b1*x+b2*x^2)
pars <- c("b1","b2")
beta <- c(3.126e-3, 2.537e-2)
cov  <- matrix(c(7.205e-06,-3.438e-06,-3.438e-06,2.718e-06),nrow=2)

ex1.a <- dose.distr(f, pars, beta, cov, cells=1811, dics=102, 
m.prior="normal", d.prior="uniform", prior.param=c(0, Inf))
summary(ex1.a)
plot(ex1.a)

---

Credible intervals for radiation dose

Description

The function allows the user to compute credible intervals for radiation doses objects of class radir.

Usage

ci.dose.radir(object, cr=0.95)

Arguments

object	the doses estimated by dose.distr function, an object of class radir.
cr	size of the credibility region. Its default value is 0.95.

Value

A vector with two elements containing the lower and upper bounds of the credible region.

Author(s)

David Moriña (Barcelona Graduate School of Mathematics), Manuel Higueras (Basque Center for Applied Mathematics) and Pedro Puig (Universitat Autònoma de Barcelona)

Mantainer: David Moriña Soler <[email protected]>

References

Higueras M, Puig P, Ainsbury EA, Rothkamm K. A new inverse regression model applied to radiation biodosimetry. Proc R Soc A 2015;471, http://dx.doi.org/10.1098/rspa.2014.0588

See Also

radir-package, dose.distr, pr.dose.radir

Examples

### Example 3 (a)
f <- expression(b1*x+b2*x^2)
pars <- c("b1","b2")
beta <- c(3.126e-3, 2.537e-2)
cov  <- matrix(c(7.205e-06,-3.438e-06,-3.438e-06,2.718e-06),nrow=2)

### (a)
ex1.a <- dose.distr(f, pars, beta, cov, cells=1811, dics=102, 
m.prior="normal", d.prior="uniform", prior.param=c(0, Inf))

ci.dose.radir(ex1.a, 0.90)

---

Inverse regression model for radiation biodosimetry

Description

The function allows the user to estimate radiation doses distribution using the methodology described in Higueras et al. (2014).

Usage

dose.distr(f, pars, beta, cov, cells, dics, m.prior="gamma", 
d.prior="uniform", prior.param=c(0,"Inf"), stdf=6, nsim=1000)

Arguments

f	dose-response function, as an expression. Must be differenciable in the domain of parameters.
pars	string vector containing the parameters in f.
beta	estimates of the parameters.
cov	variance-covariance matrix.
cells	patient information, number of cells examined.
dics	patient information, observed number of aberrations.
m.prior	string containing the prior distribution of the mean. It can be gamma (the default value) or normal.
d.prior	string containing the prior distribution of the dose. It can be gamma or uniform (the default value).
prior.param	vector of length 2 containing the parameters of the distribution of the dose prior. The parametrization for the uniform distribution is the usual, based on the support, and an improper uniform distribution is allowed, setting the second parameter to Inf. Its default value is the non-informative prior. The gamma distribution is parametrized in terms of the mean and standard deviation.
stdf	Approximated standard deviation factor. This input is useful to control the ends of the calibrative density; i.e. in case the tails of the calibrative dose density are very long this value could be reduced, or viceversa. Its default value is 6.
nsim	Number of simulations to base the results on. Its default value is 1000.

Value

An object of class dose.radir containing the distribution of the estimated doses.

Author(s)

David Moriña (Barcelona Graduate School of Mathematics), Manuel Higueras (Basque Center for Applied Mathematics) and Pedro Puig (Universitat Autònoma de Barcelona)

Mantainer: David Moriña Soler <[email protected]>

References

Higueras M, Puig P, Ainsbury EA, Rothkamm K. A new inverse regression model applied to radiation biodosimetry. Proc R Soc A 2015;471, http://dx.doi.org/10.1098/rspa.2014.0588

See Also

radir-package, ci.dose.radir, pr.dose.radir

Examples

### Example 3 (a)
f <- expression(b1*x+b2*x^2)
pars <- c("b1","b2")
beta <- c(3.126e-3, 2.537e-2)
cov  <- matrix(c(7.205e-06,-3.438e-06,-3.438e-06,2.718e-06),nrow=2)

### (a)
ex1.a <- dose.distr(f, pars, beta, cov, cells=1811, dics=102, 
m.prior="normal", d.prior="uniform", prior.param=c(0, Inf))

---

Probability between doses

Description

This function allows the user to compute the probability between two radiation doses.

Usage

pr.dose.radir(object, lod = 0, upd = object[[2]][length(object[[2]])])

Arguments

object	An object of class radir containing the estimated doses.
lod	Lower dose considered. Its default value is 0.
upd	Upper dose considered. Its default value is the maximum dose in object.

Value

The probability that the real dose is between lod and upd.

Author(s)

David Moriña (Barcelona Graduate School of Mathematics), Manuel Higueras (Basque Center for Applied Mathematics) and Pedro Puig (Universitat Autònoma de Barcelona)

Mantainer: David Moriña Soler <[email protected]>

References

Higueras M, Puig P, Ainsbury EA, Rothkamm K. A new inverse regression model applied to radiation biodosimetry. Proc R Soc A 2015;471, http://dx.doi.org/10.1098/rspa.2014.0588

See Also

radir-package, dose.distr, ci.dose.radir

Examples

### Example 3 (a)
f <- expression(b1*x+b2*x^2)
pars <- c("b1","b2")
beta <- c(3.126e-3, 2.537e-2)
cov  <- matrix(c(7.205e-06,-3.438e-06,-3.438e-06,2.718e-06),nrow=2)

### (a)
ex1.a <- dose.distr(f, pars, beta, cov, cells=1811, dics=102, 
m.prior="normal", d.prior="uniform", prior.param=c(0, Inf))

pr.dose.radir(ex1.a, 1, 1.4)

---