| Type: | Package |
| Title: | Missing Data and Measurement Error Modelling in INLA |
| Version: | 1.1.0 |
| Description: | Facilitates fitting measurement error and missing data imputation models using integrated nested Laplace approximations, according to the method described in Skarstein, Martino and Muff (2023) <doi:10.1002/bimj.202300078>. See Skarstein and Muff (2024) <doi:10.48550/arXiv.2406.08172> for details on using the package. |
| License: | MIT + file LICENSE |
| Encoding: | UTF-8 |
| LazyData: | true |
| VignetteBuilder: | knitr |
| RoxygenNote: | 7.3.1 |
| Suggests: | INLA, knitr, testthat (≥ 3.0.0), tibble, rmarkdown, spelling |
| Additional_repositories: | https://inla.r-inla-download.org/R/stable/ |
| Config/testthat/edition: | 3 |
| Imports: | dplyr, ggplot2, rlang, stats, methods, scales |
| Depends: | R (≥ 2.10) |
| URL: | https://emmaskarstein.github.io/inlamemi/, https://github.com/emmaSkarstein/inlamemi |
| Language: | en-US |
| NeedsCompilation: | no |
| Packaged: | 2024-10-31 13:42:45 UTC; emmass |
| Author: | Emma Skarstein 
[cre, aut, cph],
Stefanie Muff
[aut] |
| Maintainer: | Emma Skarstein <emma@skarstein.no> |
| Repository: | CRAN |
| Date/Publication: | 2024-10-31 16:10:02 UTC |
Extract model coefficients
Description
Extract model coefficients
Usage
## S3 method for class 'inlamemi'
coef(object, ...)
Arguments
object |
object of class 'inlamemi' |
... |
other arguments |
Extract random effects from formula
Description
Extract random effects from formula
Usage
extract_random_effects(formula)
Arguments
formula |
an object of class "formula", either the formula for the model of interest, the imputation model or the missingness model. |
Value
A list containing "reff_vars", the random effect variables, and "reff", the entire random effect term.
Extract and group variables from formulas
Description
Helper function that takes in the formulas for the model of interest and the imputation model, and groups them into responses, covariates, covariate with error and covariate(s) without error, for both sub-models.
Usage
extract_variables_from_formula(
formula_moi,
formula_imp,
formula_mis = NULL,
error_variable = NULL
)
Arguments
formula_moi |
an object of class "formula", describing the main model to be fitted. |
formula_imp |
an object of class "formula", describing the imputation model for the mismeasured and/or missing observations. |
formula_mis |
an object of class "formula", describing the missingness model. Does not need to have a response variable, since this will always be a binary missingness indicator. |
error_variable |
character vector with the name(s) of the variable(s) with error. |
Value
A list containing the names of the different variables of the model. The names of the elements in the list are "response_moi" (the response for the moi), "covariates_moi" (all covariates in the moi), "error_variable" (the name of the variable with error or missing data), "covariates_error_free" (the moi covariates without error), "response_imp" (imputation model response), "covariates_imp" (imputation model covariates).
Fit model for measurement error and missing data in INLA
Description
A wrapper function around "INLA::inla()", providing the necessary structure to fit the hierarchical measurement error model that adjusts coefficient estimates to account for biases due to measurement error and missing data.
Usage
fit_inlamemi(
formula_moi,
formula_imp = NULL,
formula_mis = NULL,
family_moi,
data,
error_type = "classical",
error_variable = NULL,
repeated_observations = FALSE,
classical_error_scaling = NULL,
prior.prec.moi = NULL,
prior.prec.berkson = NULL,
prior.prec.classical = NULL,
prior.prec.imp = NULL,
prior.beta.error = NULL,
prior.gamma.error = NULL,
initial.prec.moi = NULL,
initial.prec.berkson = NULL,
initial.prec.classical = NULL,
initial.prec.imp = NULL,
control.family.moi = NULL,
control.family.berkson = NULL,
control.family.classical = NULL,
control.family.imp = NULL,
control.family = NULL,
control.predictor = NULL,
...
)
Arguments
formula_moi |
an object of class "formula", describing the main model to be fitted. |
formula_imp |
an object of class "formula", describing the imputation model for the mismeasured and/or missing observations. |
formula_mis |
an object of class "formula", describing the missingness model. Does not need to have a response variable, since this will always be a binary missingness indicator. |
family_moi |
a string indicating the likelihood family for the model of interest (the main model). |
data |
an object of class data.frame or list containing the variables in the model. |
error_type |
type of error (one or more of "classical", "berkson", "missing") |
error_variable |
character vector with the name(s) of the variable(s) with error. |
repeated_observations |
Does the variable with measurement error and/or missingness have repeated observations? If so, set this to "TRUE". In that case, when specifying the formula, use the name of the variable without any numbers, but when specifying the data, make sure that the repeated measurements end in a number, i.e "sbp1" and "sbp2". |
classical_error_scaling |
can be specified if the classical measurement error varies across observations. Must be a vector of the same length as the data. |
prior.prec.moi |
a string containing the parameters for the prior for the precision of the residual term for the model of interest. |
prior.prec.berkson |
a string containing the parameters for the prior for the precision of the error term for the Berkson error model. |
prior.prec.classical |
a string containing the parameters for the prior for the precision of the error term for the classical error model. |
prior.prec.imp |
a string containing the parameters for the precision of the latent variable x, which is the variable being described in the imputation model. |
prior.beta.error |
parameters for the Gaussian prior for the coefficient of the error prone variable. |
prior.gamma.error |
parameters for the Gaussian prior for the coefficient of the variable with missingness in the missingness model. |
initial.prec.moi |
the initial value for the precision of the residual term for the model of interest. |
initial.prec.berkson |
the initial value for the precision of the residual term for the Berkson error term. |
initial.prec.classical |
the initial value for the precision of the residual term for the classical error term. |
initial.prec.imp |
the initial value for the precision of the residual term for the latent variable r. |
control.family.moi |
control.family component for model of interest. Can be specified here using the inla syntax instead of passing the "prior.prec..." and "initial.prec..." arguments, or in the cases when other hyperparameters are needed for the model of interest, see for instance survival models. |
control.family.berkson |
control.family component Berkson model. Can be specified here using the inla syntax instead of passing the "prior.prec..." and "initial.prec..." arguments. Useful in the cases when more flexibility is needed, for instance if one wants to specify a different prior distribution than Gamma. |
control.family.classical |
control.family component for classical model. Can be specified here using the inla syntax instead of passing the "prior.prec..." and "initial.prec..." arguments. Useful in the cases when more flexibility is needed, for instance if one wants to specify a different prior distribution than Gamma. |
control.family.imp |
control.family component for imputation model. Can be specified here using the inla syntax instead of passing the "prior.prec..." and "initial.prec..." arguments. Useful in the cases when more flexibility is needed, for instance if one wants to specify a different prior distribution than Gamma. |
control.family |
control.family for use in inla (can be provided directly instead of passing the "prior.prec...." and "initial.prec..." arguments. If this is specified, any other "control.family..." or "prior.prec..." arguments provided will be ignored. |
control.predictor |
control.predictor for use in inla. |
... |
other arguments to pass to 'inla'. |
Value
An object of class inlamemi.
Examples
# Fit the model
simple_model <- fit_inlamemi(data = simple_data,
formula_moi = y ~ x + z,
formula_imp = x ~ z,
family_moi = "gaussian",
error_type = c("berkson", "classical"),
error_variable = "x",
prior.prec.moi = c(10, 9),
prior.prec.berkson = c(10, 9),
prior.prec.classical = c(10, 9),
prior.prec.imp = c(10, 9),
prior.beta.error = c(0, 1/1000),
initial.prec.moi = 1,
initial.prec.berkson = 1,
initial.prec.classical = 1,
initial.prec.imp = 1)
Framingham heart study data
Description
A data set with observations of heart disease status systolic blood pressure (SBP) and smoking status.
Usage
framingham
Format
## 'framingham' A data frame with 641 rows and 4 columns:
- disease
A binary response, 1 if heart disease, 0 otherwise
- sbp1
log(SBP - 50) at examination 1 (centered)
- sbp2
log(SBP - 50) at examination 2 (centered)
- smoking
Smoking status, 1 if smoking, 0 otherwise.
Source
MacMahon et al. (1990) <https://doi.org/10.1016/0140-6736(90)90878-9>
Extract coefficients for the imputation model (IMP)
Description
Extract coefficients for the imputation model (IMP)
Usage
get_coef_imp(inlamemi_model)
Arguments
inlamemi_model |
object of class 'inlamemi' |
Value
A data frame with a summary of the posterior marginals for the coefficients in the imputation model.
Extract coefficients for the missingness model (MIS)
Description
Extract coefficients for the missingness model (MIS)
Usage
get_coef_mis(inlamemi_model)
Arguments
inlamemi_model |
object of class 'inlamemi' |
Value
A data frame with a summary of the posterior marginals for the coefficients in the missingness model.
Extract coefficients for the model of interest (MOI)
Description
Extract coefficients for the model of interest (MOI)
Usage
get_coef_moi(inlamemi_model)
Arguments
inlamemi_model |
object of class 'inlamemi' |
Value
A data frame with a summary of the posterior marginals for the coefficients in the model of interest.
Extract imputed values
Description
Extract imputed values
Usage
get_imputed(inlamemi_model, error_variable)
Arguments
inlamemi_model |
object of class 'inlamemi' |
error_variable |
character string indicating the name of the error variable for which to extract the imputed values. |
Value
A list of two objects: the posterior marginal distributions for each element of the imputed covariate, and a data frame giving a summary of these marginals.
List the survival likelihoods in INLA
Description
List the survival likelihoods in INLA
Usage
inla_survival_families()
Value
List of survival models in INLA
Make "control.family" argument for passing to the "inla" function
Description
Make "control.family" argument for passing to the "inla" function
Usage
make_inlamemi_control.family(
formula_mis = NULL,
family_moi,
error_type = "classical",
prior.prec.moi = NULL,
prior.prec.berkson = NULL,
prior.prec.classical = NULL,
prior.prec.imp = NULL,
initial.prec.moi = NULL,
initial.prec.berkson = NULL,
initial.prec.classical = NULL,
initial.prec.imp = NULL,
control.family.moi = NULL,
control.family.berkson = NULL,
control.family.classical = NULL,
control.family.imp = NULL,
control.family = NULL
)
Arguments
formula_mis |
an object of class "formula", describing the missingness model. Does not need to have a response variable, since this will always be a binary missingness indicator. |
family_moi |
a string indicating the likelihood family for the model of interest (the main model). |
error_type |
type of error (one or more of "classical", "berkson", "missing") |
prior.prec.moi |
a string containing the parameters for the prior for the precision of the residual term for the model of interest. |
prior.prec.berkson |
a string containing the parameters for the prior for the precision of the error term for the Berkson error model. |
prior.prec.classical |
a string containing the parameters for the prior for the precision of the error term for the classical error model. |
prior.prec.imp |
a string containing the parameters for the precision of the latent variable x, which is the variable being described in the imputation model. |
initial.prec.moi |
the initial value for the precision of the residual term for the model of interest. |
initial.prec.berkson |
the initial value for the precision of the residual term for the Berkson error term. |
initial.prec.classical |
the initial value for the precision of the residual term for the classical error term. |
initial.prec.imp |
the initial value for the precision of the residual term for the latent variable r. |
control.family.moi |
control.family component for model of interest. Can be specified here using the inla syntax instead of passing the "prior.prec..." and "initial.prec..." arguments, or in the cases when other hyperparameters are needed for the model of interest, see for instance survival models. |
control.family.berkson |
control.family component Berkson model. Can be specified here using the inla syntax instead of passing the "prior.prec..." and "initial.prec..." arguments. Useful in the cases when more flexibility is needed, for instance if one wants to specify a different prior distribution than Gamma. |
control.family.classical |
control.family component for classical model. Can be specified here using the inla syntax instead of passing the "prior.prec..." and "initial.prec..." arguments. Useful in the cases when more flexibility is needed, for instance if one wants to specify a different prior distribution than Gamma. |
control.family.imp |
control.family component for imputation model. Can be specified here using the inla syntax instead of passing the "prior.prec..." and "initial.prec..." arguments. Useful in the cases when more flexibility is needed, for instance if one wants to specify a different prior distribution than Gamma. |
control.family |
control.family for use in inla (can be provided directly instead of passing the "prior.prec...." and "initial.prec..." arguments. If this is specified, any other "control.family..." or "prior.prec..." arguments provided will be ignored. |
Value
the "control.family" argument to be passed to inla, a list of "control.family" arguments for each model in the hierarchical measurement error model.
Examples
make_inlamemi_control.family(
family_moi = "gaussian",
error_type = c("berkson", "classical"),
prior.prec.moi = c(10, 9),
prior.prec.berkson = c(10, 9),
prior.prec.classical = c(10, 9),
prior.prec.imp = c(10, 9),
initial.prec.moi = 1,
initial.prec.berkson = 1,
initial.prec.classical = 1,
initial.prec.imp = 1)
make_inlamemi_control.family(
family_moi = "weibull.surv",
error_type = c("classical", "missing"),
control.family.moi =
list(hyper = list(alpha = list(param = 0.01,
initial = log(1.4),
fixed = FALSE))),
prior.prec.classical = c(0.5, 0.5),
prior.prec.imp = c(0.5, 0.5),
initial.prec.classical = 2.8,
initial.prec.imp = 1)
Make vector of likelihood families
Description
Make vector of likelihood families
Usage
make_inlamemi_families(family_moi, inlamemi_stack)
Arguments
family_moi |
a string indicating the likelihood family for the model of interest (the main model). |
inlamemi_stack |
object of type inla.stack |
Value
A vector specifying the likelihood family for each model level.
Examples
simple_stack <- make_inlamemi_stacks(formula_moi = y ~ x + z,
formula_imp = x ~ z,
data = simple_data,
error_type = c("classical"))
make_inlamemi_families(family_moi = "gaussian",
inlamemi_stack = simple_stack)
Make formula for measurement error and missing data model
Description
Make formula for measurement error and missing data model
Usage
make_inlamemi_formula(
formula_moi,
formula_imp,
formula_mis = NULL,
family_moi = "gaussian",
error_type = "classical",
error_variable = NULL,
prior.beta.error,
prior.gamma.error = NULL,
vars = NULL
)
Arguments
formula_moi |
an object of class "formula", describing the main model to be fitted. |
formula_imp |
an object of class "formula", describing the imputation model for the mismeasured and/or missing observations. |
formula_mis |
an object of class "formula", describing the missingness model. Does not need to have a response variable, since this will always be a binary missingness indicator. |
family_moi |
a string indicating the likelihood family for the model of interest (the main model). |
error_type |
type of error (one or more of "classical", "berkson", "missing") |
error_variable |
character vector with the name(s) of the variable(s) with error. |
prior.beta.error |
parameters for the Gaussian prior for the coefficient of the error prone variable. |
prior.gamma.error |
parameters for the Gaussian prior for the coefficient of the variable with missingness in the missingness model. |
vars |
Results from a call to "extract_variables_from_formula" function. If this is not passed as an argument, it is called inside the function. |
Value
An object of class "formula".
Examples
make_inlamemi_formula(formula_moi = y ~ x + z,
formula_imp = x ~ z,
error_type = "classical",
prior.beta.error = c(0, 1/1000)
)
Construct scaling vector to scale the precision of correctly observed observations
Description
Construct scaling vector to scale the precision of correctly observed observations
Usage
make_inlamemi_scaling_vector(
inlamemi_stack,
error_type,
classical_error_scaling = NULL,
vars
)
Arguments
inlamemi_stack |
an object of class inlamemi.data.stack containing data structured |
error_type |
type of error (one or more of "classical", "berkson", "missing") |
classical_error_scaling |
can be specified if the classical measurement error varies across observations. Must be a vector of the same length as the data. |
vars |
Results from a call to "extract_variables_from_formula" function. If this is not passed as an argument, it is called inside the function. |
Value
A vector reflecting the scaling factor for the residual terms in each model level.
Examples
stacks <- make_inlamemi_stacks(data = simple_data,
formula_moi = y ~ x + z,
formula_imp = x ~ z,
error_type = c("classical", "berkson"))
vars <- extract_variables_from_formula(formula_moi = y ~ x + z,
formula_imp = x ~ z)
make_inlamemi_scaling_vector(stacks,
error_type = c("classical", "berkson"),
vars = vars)
Make data stacks for joint model specification in INLA
Description
Make data stacks for joint model specification in INLA
Usage
make_inlamemi_stacks(
formula_moi,
formula_imp,
formula_mis = NULL,
family_moi = "gaussian",
data,
error_type = "classical",
error_variable = NULL,
repeated_observations = FALSE,
vars = NULL
)
Arguments
formula_moi |
an object of class "formula", describing the main model to be fitted. |
formula_imp |
an object of class "formula", describing the imputation model for the mismeasured and/or missing observations. |
formula_mis |
an object of class "formula", describing the missingness model. Does not need to have a response variable, since this will always be a binary missingness indicator. |
family_moi |
a string indicating the likelihood family for the model of interest (the main model). |
data |
an object of class data.frame or list containing the variables in the model. |
error_type |
type of error (one or more of "classical", "berkson", "missing") |
error_variable |
character vector with the name(s) of the variable(s) with error. |
repeated_observations |
Does the variable with measurement error and/or missingness have repeated observations? If so, set this to "TRUE". In that case, when specifying the formula, use the name of the variable without any numbers, but when specifying the data, make sure that the repeated measurements end in a number, i.e "sbp1" and "sbp2". |
vars |
Results from a call to "extract_variables_from_formula" function. If this is not passed as an argument, it is called inside the function. |
Value
An object of class inla.stack with data structured according to specified formulas and error models.
Examples
make_inlamemi_stacks(formula_moi = y ~ x + z,
formula_imp = x ~ z,
data = simple_data,
error_type = "classical")
Simulated data with observation missing at random (MAR)
Description
A simulated dataset to demonstrate how to set up a model in the case where there are two variables with measurement error.
Usage
mar_data
Format
## 'mar_data' A data frame with 1000 rows and 5 columns:
- y
Response variable
- x
Observed value of covariate, with almost 20 percent missing
- x_true
Correct version of x, without missingness
- z1
Covariate correlated with x
- z2
Covariate correlated with the missingness of x
Source
The dataset is simulated.
Survival data with repeated systolic blood pressure measurements
Description
A dataset containing a repeated blood pressure measurement along with some other variables for participants in the Third National Health and Nutrition Survey (NHANES III), merged with data from the US National Death Index by Ruth H. Keogh and Jonathan Bartlett. For the illustration purposes in this package, we have left out observations where smoking status is missing.
Usage
nhanes_survival
Format
## 'nhanes_survival' A data frame with 3433 rows and 8 columns:
- sbp1
systolic blood pressure (standardized), first measurement
- sbp2
systolic blood pressure (standardized), second measurement
- sex
sex (0 = female, 1 = male)
- age
age (standardized)
- smoke
smoking status (0 = no, 1 = yes)
- diabetes
diabetes status (0 = no, 1 = yes)
- d
censoring status (0 = censored, 1 = observed death due to cardiovascular disease)
- t
time until death due to cardiovascular disease occurs
Source
https://github.com/ruthkeogh/meas_error_handbook
Plot model summary
Description
Plot model summary
Usage
## S3 method for class 'inlamemi'
plot(
x,
plot_moi = TRUE,
plot_imp = TRUE,
plot_mis = TRUE,
plot_intercepts = TRUE,
error_variable_highlight = FALSE,
greek = FALSE,
palette = NULL,
...
)
Arguments
x |
the model returned from the fit_inlamemi function. |
plot_moi |
should the posterior mean for the coefficients of the model of interest be plotted? Defaults to TRUE. |
plot_imp |
should the posterior mean for the coefficients of the imputation model be plotted? Defaults to TRUE. |
plot_mis |
should the posterior mean for the coefficients of the missingness model be plotted? Defaults to TRUE. |
plot_intercepts |
should the posterior mean for the intercept(s) be plotted? Defaults to TRUE. |
error_variable_highlight |
should the coefficient(s) of the variable(s) with error be highlighted? (circled in black) Defaults to FALSE. |
greek |
make the coefficient names into greek letters with the covariate name as subscript. Defaults to FALSE. |
palette |
either a number (between 1 and 5), indicating the number of the color palette to be used, or a vector of the colors to be used. |
... |
other arguments |
Value
An object of class "ggplot2" that plots the posterior mean and 95 % credible interval for each coefficient in the model. The coefficients are colored to indicate if they belong to the main or imputation model, and the variable with error is also highlighted.
Examples
# Fit the model
simple_model <- fit_inlamemi(data = simple_data,
formula_moi = y ~ x + z,
formula_imp = x ~ z,
family_moi = "gaussian",
error_type = c("berkson", "classical"),
prior.prec.moi = c(10, 9),
prior.prec.berkson = c(10, 9),
prior.prec.classical = c(10, 9),
prior.prec.imp = c(10, 9),
prior.beta.error = c(0, 1/1000),
initial.prec.moi = 1,
initial.prec.berkson = 1,
initial.prec.classical = 1,
initial.prec.imp = 1)
plot(simple_model)
Print method for inlamemi
Description
Print method for inlamemi
Usage
## S3 method for class 'inlamemi'
print(x, ...)
Arguments
x |
object of class 'inlamemi'. |
... |
other arguments. |
Visualize the model data structure as matrices in LaTeX
Description
Visualize the model data structure as matrices in LaTeX
Usage
show_data_structure(stack)
Arguments
stack |
an object of class inla.stack returned from the function make_inlamemi_stacks, which describes the structure of the data for the measurement error and imputation model. |
Value
A list containing data frames with the left hand side (response_df) and right hand side (effects_df), along with the latex code needed to visualize the matrices (matrix_string).
Examples
stack <- make_inlamemi_stacks(data = simple_data,
formula_moi = y ~ x + z,
formula_imp = x ~ z,
error_type = "classical")
show_data_structure(stack)
Simple simulated data
Description
A simulated dataset to demonstrate how to model different types of measurement error and missing data using the 'inlamemi' package.
Usage
simple_data
Format
## 'simple_data' A data frame with 1000 rows and 4 columns:
- y
Response variable
- x
Covariate measured with error, both Berkson and classical error and missing observations
- x_true
Correct version of the covariate with error
- z
Error free covariate, correlated with x
Source
The dataset is simulated.
Simplify the "raw" model summary for printing and plotting
Description
Simplify the "raw" model summary for printing and plotting
Usage
simplify_inlamemi_model_summary(inlamemi_model)
Arguments
inlamemi_model |
the model returned from the fit_inlamemi function. |
Value
A list of four data frames, containing the summaries for different components of the model. These are the coefficients of the model of interest, the coefficient of the variable with error, the coefficients of the imputation model, and the hyperparameters.
Summary method for inlamemi
Description
Takes a fitted 'inlamemi' object produced by 'fit_inlamemi' and produces a summary from it.
Usage
## S3 method for class 'inlamemi'
summary(object, ...)
## S3 method for class 'summary.inlamemi'
print(x, ...)
Arguments
object |
model of class 'inlamemi'. |
... |
other arguments |
x |
object of class summary.inlamemi. |
Value
'summary.inlamemi' returns an object of class 'summary.inlamemi', a list of components to print.
Examples
# Fit the model
simple_model <- fit_inlamemi(data = simple_data,
formula_moi = y ~ x + z,
formula_imp = x ~ z,
family_moi = "gaussian",
error_type = c("berkson", "classical"),
prior.prec.moi = c(10, 9),
prior.prec.berkson = c(10, 9),
prior.prec.classical = c(10, 9),
prior.prec.imp = c(10, 9),
prior.beta.error = c(0, 1/1000),
initial.prec.moi = 1,
initial.prec.berkson = 1,
initial.prec.classical = 1,
initial.prec.imp = 1)
summary(simple_model)
Simulated data with two covariates with classical measurement error
Description
A simulated dataset to demonstrate how to set up a model in the case where there are two variables with measurement error.
Usage
two_error_data
Format
## 'two_error_data' A data frame with 1000 rows and 5 columns:
- y
Response variable
- x1
Covariate measured with classical error, correlated with z
- x2
Covariate measured with classical error
- x1_true
Correct version of x1
- x2_true
Correct version of x2
- z
Error free covariate, correlated with x1
Source
The dataset is simulated.