nca.pde.deviation.outlier function

Calculates individual prediction distribution errors (PDE) and scaled deviation of NCA metrics estimated from observed and simulated data. Identifies outlier to population PK model.

nca.pde.deviation.outlier calculates individual prediction distribution errors (PDE) and scaled deviation of NCA metrics estimated from observed and simulated data. Identifies outlier to population PK model.

nca.pde.deviation.outlier(
  obsdata,
  simdata,
  idNm = "ID",
  id = NULL,
  spread = "npi",
  figlbl = NULL,
  calcparam = c("AUClast", "Cmax"),
  diagparam = c("AUClast", "Cmax"),
  cunit = NULL,
  tunit = NULL,
  noPlot = FALSE,
  onlyNCA = onlyNCA
)

Arguments

• obsdata: A data frame containing the NCA metrics values estimated from the observed data
• simdata: A data frame containing the NCA metrics values estimated from the simulated data
• idNm: Column name for ID ("ID" )
• id: ID of the individual whose data is being evaluated
• spread: Measure of the spread of simulated data (ppi (95% parametric prediction interval) or npi (95% nonparametric prediction interval)) ("npi" )
• figlbl: Figure label based on dose identifier and/or population stratifier, in addition to ID (NULL )
• calcparam: A character array of the NCA metrics used for calculations of PDE and deviation. The allowed NCA metrics for this histograms are "AUClast", "AUClower_upper", "AUCINF_obs", "AUCINF_pred", "AUMClast", "Cmax", "Tmax" and "HL_Lambda_z". (c("AUClast", "Cmax") )
• diagparam: A character array of the NCA metrics used for diagnostic test to detect outliers. The allowed NCA metrics for this histograms are "AUClast", "AUClower_upper", "AUCINF_obs", "AUCINF_pred", "AUMClast", "Cmax", "Tmax" and "HL_Lambda_z". (c("AUClast", "Cmax") )
• cunit: Unit for concentration (default is ‘NULL’ )
• tunit: Unit for time (default is ‘NULL’ )
• noPlot: Perform only NCA calculations without any plot generation (TRUE, FALSE) (FALSE )
• onlyNCA: If TRUE only NCA is performed and ppc part is ignored although simFile is not NULL. Default is ‘FALSE’

Returns

returns the observed data frame with added distance and simulation mean of the nCA metrics, and a data frame with the PDE values of the NCA metrics. If the individual is identified as an outlier for the PK model, histograms of the diagnostic NCA metrics are produced and a graphical object created by arrangeGrob function is returned.

Details

nca.pde.deviation.outlier calculates individual prediction distribution errors (PDE) and scaled deviation of NCA metrics estimated from observed and simulated data. The deviation of each estimated NCA metrics is scaled by the "spread" of the simulated values. The "spread" is measured either by the 95% parametric prediction interval or 95% non-parametric prediction interval. Any individual yielding an absolute value of the scaled deviation for any of the selected NCA metrics greater than 1, is assigned as an outlier to the corresponding population PK model. The allowed NCA metrics for this diagnostic tests are "AUClast", "AUClower_upper", "AUCINF_obs", "AUCINF_pred", "AUMClast", "Cmax", "Tmax" and "HL_Lambda_z". By default, this function uses AUClast and Cmax metrics for the comparison.