Skip to contents

`r lifecycle::badge("stable")`

NPparse processes the output from an NPAG run into a list.

Usage

NPparse(outfile = "NP_RF0001.TXT")

Arguments

outfile

This is the filename of the output from NPAG. Typically, the file will be called NP_RF0001.txt, and this is the default.

Value

The output of NPparse is a list with the following objects and of the class NPAG.

nsub

Number of subjects

nactve

Number of active grid points at the final cycle

nvar

Number of random variables or parameters in the model

nofix

Number of fixed variables or parameters in the model

par

Names of random parameters

parfix

Names of fixed parameters

covnames

Names of covariates

ab

Initial boundaries for each random parameter

valfix

Values for fixed parameters

ndim

Number of differential equations in model, or 0 for only output equation, or -1 for analytic solution (algabraic)

indpts

Index for the initial number of gridpoints in the model

icycst

Starting cycle number

icycmax

Maximum number of cycles specified by the user

icyctot

Number of cycles run. If less than icycmax, convergence occurred.

converge

Boolean value if convergence occurred.

ODEtol

Ordindary Differential Equation solver tolerance.

prior

Prior density for the run, either “UNIFORM” or the name of the user-specified density file, typically “DEN0001”.

ERRmod

Assay error model: 1 for SD; 2 for \(SD*gamma\); 3 for additive lambda model; and 4 for gamma only

numeqt

Number of output equations

ndrug

Number of drug inputs

salt

Vector of values of the salt fraction for each ndrug

ndose

Vector of the number of doses for each subject in the population

ncov

Number of covariates in the model

nobs

Vector of the number of observations for each subject in the population

nobsmax

Maximum number of observation in any individual subject

numt

Vector of the number of time points for each subject at which a prediction is generated for each numeqt output equation

corden

Final cycle joint population density of parameter estimates

postden

Array of posterior parameter value distributions for the first 100 subjects at each observation time point. postden[nsub,nactvepost,density] where nactvepost is the posterior grid point

pyjgx

Matrix of posterior probability of each nactve point for each subject, given that subject's data

ypredpop

Array of population model predictions for each subject at each observation time point. ypredpop[nsub,numeqt,time,type] where type is 1=mean, 2=median, 3=mode of the population prior used to calculate ypredpop

ypredbay

Array of Bayesian posterior model predictions for each subject at each observation time point. ypredbay[nsub,numeqt,time,type] where type is 1=mean, 2=median, 3=mode of the population prior used to calculate ypredbay

ttpred

Matrix of the prediction time points for each subject, with nsub rows and max(numt) columns

exx

Array of the mean, median, and mode of the posterior marginal distribution for each parameter in each subject, of the form exx[nvar,type,nsub]

ypredpopt

Array of population model predictions for each subject at each ttpred time point, of the form ypredpopt[nsub, numeqt, time, type], where type is 1=mean, 2=median, 3=mode of the population prior used to calculate ypredpopt

ilog

Matrix of cycle number and associated log-likelihood

iic

Matrix with cycle number and Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) for each cycle

imean

Matrix of cycle numbers and associated means for each random parameter

isd

Matrix of cycle numbers and associated standard deviations for each random parameter

iaddl

Array of additional information for each random parameter in each cycle, of the form iaddl[info, nvar, cycle], where info is a value from 1 to 12: 1= mode; 2= skewness; 3= kurtosis; 4-8 give percentiles of the distribution where 4=2.5%; 5=25%; 6=50% (median); 7=75%; 8=97.5%; 9= the standard deviation of a normal distribution with the same interquartile range; 10=the standard deviation of a normal distribution with the same 95% range; 11=the average of 9 and 10; 12=the % scaled information

igamlam

Matrix of cycle number and associated gamma or lambda

blog

Vector of each subject's Bayesian posterior log-likelihood

bmean

Matrix of subject numbers and associated Bayesian posterior means for each random parameter

bsd

Matrix of subject numbers and associated Bayesian posterior standard deviations for each random parameter

baddl

Array of Bayesian posterior additional information for each random parameter for each subject, of the form baddl[info, nvar, nsub], where info is the same as for iaddl.

bauc

Matrix of AUC blocks for each subject with 5 columns: [nsub, numeqt, nblock, tau, auc]; nsub and numeqt are as previously defined; nblock is the AUC block as defined by successive dose reset (evid=4) events; tau is the time interval for that block; auc is the AUC for that block

sdata

Subject data consisting of 5 columns: [id, nsub, age, sex, ht], id is the original identification number in the .csv matrix file; nsub is the sequential subject number in the NPAG run; age, sex and ht will be missing for .csv input and present if included in .wrk input files

dosecov

Matrix with all dosing information for each subject, including times, routes, amounts, and associated covariate values

outputs

Matrix with measured outputs for each subject and associated assay error polynomials. The order of the columns is nsub, time, numeqt, observation, c0, c1, c2, c3, where the last four columns are the coefficients of the assay error polynomial for that observation, such that \(SD[obs] = c0 + c1*[obs] + c2*[obs]**2 + c3*[obs]**3\)

negflag

A flag indicating that some negative predictions were changed to missing. This means that the model may be misspecified.

mdata

The filename of the data used in the run.

Details

This function is called automatically at the end of a run. It can take some time to complete, depending on the number of subjects, doses, observations, etc. Typical wait times are a few seconds up to 5 minutes. When processing is complete a summary of the extracted data will be reported on the console.

Author

Michael Neely