Package 'RESIDE'

RESIDE: Rapid Easy Synthesis to Inform Data Extraction

Description

Assists researchers with planning analysis prior to obtaining data from Trusted Research Environments (TREs), also known as safe havens. Marginal distributions of one or more related data frames can be exported from a TRE and imported elsewhere, where data can be synthesised from them, with or without user specified correlations, by sampling from a multivariate cumulative distribution (copula). The International Stroke Trial (IST) is included as an example dataset under the ODC-By licence, Sandercock et al. (2011) doi:10.7488/ds/104, Sandercock et al. (2011) doi:10.1186/1745-6215-12-101.

Details

The RESIDE Package

This work was supported by the UKRI Strength in Places Fund (SIPF) Competition, #' project number 107140. The project title is SIPF The Living Laboratory driving economic growth in Glasgow through real world implementation of precision medicine.

Author(s)

Maintainer: Ryan Field [email protected] (ORCID)

Authors:

• Ryan Field [email protected] (ORCID)

• David McAllister [email protected] (ORCID)

Other contributors:

• Claudia Geue [email protected] (ORCID) [contributor]

See Also

Useful links:

• https://hehta.github.io/RESIDE/

---

Create a correlation object

Description

A helper function to create a correlation object

Usage

correlation(x, y, rho, ...)

Arguments

x	The name of the first variable
y	The name of the second variable
rho	The correlation between the two variables
...	Additional arguments to specify data frame names and factor names See details for more information on the additional arguments.

Details

This function is a helper function to create a correlation object that can be used to specify correlations between variables when synthesising data using the synthesise_data function. Additional Arguments:

• df_name: The name of the data frame containing both variables

• df_name.x: The name of the data frame containing the first variable

• df_name.y: The name of the data frame containing the second variable

• factor_name.x: The name of the factor variable for the first variable

• factor_name.y: The name of the factor variable for the second variable

Value

A list containing the correlation information

Examples

correlation("age", "bmi", 0.5)

---

Export an empty correlation matrix (removed)

Description

This function has been removed. Correlations should now be specified using the correlation function.

Usage

export_empty_cor_matrix(...)

Arguments

...	Ignored, retained for backwards compatibility.

Details

Previously this function exported an empty correlation matrix as a csv file. Correlations are now supplied directly to synthesise_data as a list of objects created with the correlation function.

Value

No return value, always throws an error.

See Also

correlation

Examples

try(export_empty_cor_matrix())

---

Export Marginal Distributions

Description

Export the marginal distributions to CSV files

Usage

export_marginal_distributions(
  marginals,
  folder_path,
  create_folder = FALSE,
  force = FALSE
)

Arguments

marginals	an Object of type RESIDE from get_marginal_distributions
folder_path	path to folder where to save files.
create_folder	if the folder does not exist should it be created, Default: FALSE
force	if the folder already contains marginal distribution files should they be removed, Default: FALSE

Details

Exports each of the marginal distributions to CSV files within a given folder, along with the continuous quantiles.

Value

No return value, called for exportation of files.

See Also

get_marginal_distributions

Examples

marginal_distributions <- get_marginal_distributions(
  IST,
  variables = c("SEX", "AGE", "RSBP", "RATRIAL")
)
export_marginal_distributions(
  marginal_distributions,
  folder_path = file.path(tempdir(), "marginals"),
  create_folder = TRUE,
  force = TRUE
)

---

Generate Marginal Distributions for a given data frame

Description

Generate Marginal Distributions from a given data frame with options to specify which variables to use.

Usage

get_marginal_distributions(
  df,
  subject_identifier = "",
  variables = c(),
  print = FALSE,
  retype = TRUE
)

Arguments

df	Data frame or a "list" of data frames to get the marginal distributions from
subject_identifier	(Optional) Subject identifier required if a list of data frames is provided, Default: ""
variables	(Optional) variable (columns) to select, Default: c()
print	Whether to print the marginal distributions to the console, Default: FALSE
retype	Whether to re-type the data frame, Default: TRUE

Details

A function to generate marginal distributions from a given data frame, depending on the variable type the marginals will differ, for binary variables a mean and number of missing is generated for continuous variables, they are first transformed and both mean and sd of the transformed variables are stored along with the quantile mapping for back transformation. For categorical variables, the number of each category is stored, missing values are categorise as "missing".

Value

A list of marginal distributions of an S3 RESIDE Class

See Also

export_marginal_distributions

Examples

marginal_distributions <- get_marginal_distributions(
  IST,
  variables = c(
    "SEX",
    "AGE",
    "ID14",
    "RSBP",
    "RATRIAL"
  )
)

---

Import a correlation matrix (removed)

Description

This function has been removed. Correlations should now be specified using the correlation function.

Usage

import_cor_matrix(...)

Arguments

...	Ignored, retained for backwards compatibility.

Details

Previously this function imported a correlation matrix from a csv file. Correlations are now supplied directly to synthesise_data as a list of objects created with the correlation function.

Value

No return value, always throws an error.

See Also

correlation

Examples

try(import_cor_matrix())

---

Import Marginal Distributions

Description

Import the marginal distribution as exported from a Trusted Research Environment (TRE)

Usage

import_marginal_distributions(
  folder_path = ".",
  binary_variables_file = "",
  categorical_variables_file = "",
  continuous_variables_file = "",
  summary_file = "summary.csv"
)

Arguments

folder_path	Where the marginal distribution files are located, Default: '.' see details.
binary_variables_file	filename for the binary_variables file, Default: ” see details.
categorical_variables_file	filename for the categorical variables file , Default: ” see details.
continuous_variables_file	filename for the continuous variables file, Default: ” see details.
summary_file	filename for the summary file, Default: 'summary.csv' see details.

Details

This function will import marginal distributions as generated within a Trusted Research Environment (TRE) using the function export_marginal_distributions. The folder_path allows the path of the files provided by the TRE to be imported, this will default to the current working directory. The file parameters will provide the default file names if no filenames are specified.

Value

Returns an object of a RESIDE class

See Also

synthesise_data

Examples

# Export marginal distributions to a temporary folder
folder_path <- file.path(tempdir(), "marginals")
export_marginal_distributions(
  get_marginal_distributions(
    IST,
    variables = c("SEX", "AGE", "RSBP", "RATRIAL")
  ),
  folder_path = folder_path,
  create_folder = TRUE,
  force = TRUE
)
# Import the marginal distributions
marginals <- import_marginal_distributions(folder_path = folder_path)

---

IST Dataset

Description

The International Stroke Trial Dataset

Usage

IST

Format

A data frame with 19435 rows and 112 columns:

AGE

Randomisation data: Age in years

CMPLASP

Other data and derived variables: Compliant for aspirin

CMPLHEP

Other data and derived variables: Compliant for heparin

CNTRYNUM

Other data and derived variables: Country code

COUNTRY

Other data and derived variables: Abbreviated country code

DALIVE

Recurrent stroke within 14 days: Discharged alive from hospital

DALIVED

Recurrent stroke within 14 days: Date Discharged alive from hospital

DAP

Data collected on 14 day/discharge form about treatments given in hospital: Non trial antiplatelet drug (Y/N)

DASP14

Data collected on 14 day/discharge form about treatments given in hospital: Aspirin given for 14 days or till death or discharge (Y/N)

DASPLT

Data collected on 14 day/discharge form about treatments given in hospital: Discharged on long term aspirin (Y/N)

DAYLOCAL

Randomisation data: Estimate of local day of week (assuming RDATE is Oxford)

DCAA

Data collected on 14 day/discharge form about treatments given in hospital: Calcium antagonists (Y/N)

DCAREND

Data collected on 14 day/discharge form about treatments given in hospital: Carotid surgery (Y/N)

DDEAD

Other events within 14 days: Dead on discharge form

DDEADC

Other events within 14 days: Cause of death (1-Initial stroke/2-Recurrent stroke (ischaemic or unknown /3-Recurrent stroke (haemorrhagic)/4-Pneumonia /5-Coronary heart disease/6-Pulmonary embolism /7-Other vascular or unknown/8-Non-vascular/0-unknown)

DDEADD

Date of dead on discharge form (yyyy/mm/dd); NOTE: this death is not necessarily within 14 days of randomisation

DDEADX

Other events within 14 days: Comment on death

DDIAGHA

Final diagnosis of initial event: Haemorrhagic stroke

DDIAGISC

Final diagnosis of initial event: Ischaemic stroke

DDIAGUN

Final diagnosis of initial event: Indeterminate stroke

DEAD1

Indicator variables for specific causes of death: Initial stroke

DEAD2

Indicator variables for specific causes of death: Reccurent ischaemic/unknown stroke

DEAD3

Indicator variables for specific causes of death: Reccurent haemorrhagic stroke

DEAD4

Indicator variables for specific causes of death: Pneumonia

DEAD5

Indicator variables for specific causes of death: Coronary heart disease

DEAD6

Indicator variables for specific causes of death: Pulmonary embolism

DEAD7

Indicator variables for specific causes of death: Other vascular or unknown

DEAD8

Indicator variables for specific causes of death: Non vascular

DGORM

Data collected on 14 day/discharge form about treatments given in hospital: Glycerol or manitol (Y/N)

DHAEMD

Data collected on 14 day/discharge form about treatments given in hospital: Haemodilution (Y/N)

DHH14

Data collected on 14 day/discharge form about treatments given in hospital: Medium dose heparin given for 14 days etc in pilot (combine with above)

DIED

Other data and derived variables: Indicator variable for death (1=died; 0=did not die)

DIVH

Data collected on 14 day/discharge form about treatments given in hospital: Non trial intravenous heparin (Y/N)

DLH14

Data collected on 14 day/discharge form about treatments given in hospital: Low dose heparin given for 14 days or till death/discharge (Y/N)

DMAJNCH

Data collected on 14 day/discharge form about treatments given in hospital: Major non-cerebral haemorrhage (Y/N)

DMAJNCHD

Data collected on 14 day/discharge form about treatments given in hospital: Date of Major non-cerebral haemorrhage (yyyy/mm/dd)

DMAJNCHX

Data collected on 14 day/discharge form about treatments given in hospital: Comment of Major non-cerebral haemorrhage

DMH14

Data collected on 14 day/discharge form about treatments given in hospital: Date of Major non-cerebral haemorrhage (yyyy/mm/dd)

DNOSTRK

Final diagnosis of initial event: Not a stroke

DNOSTRKX

Final diagnosis of initial event: Comment on Not a stroke

DOAC

Data collected on 14 day/discharge form about treatments given in hospital: Other anticoagulants (Y/N)

DPE

Other events within 14 days: Pulmonary embolism

DPED

Other events within 14 days: Date of Pulmonary embolism (yyyy/mm/dd)

DPLACE

Other events within 14 days: Discharge destination (A-Home /B-Relatives home /C-Residential care /D-Nursing home /E-Other hospital departments /U-Unknown)

DRSH

Recurrent stroke within 14 days: Haemorrhagic stroke

DRSHD

Recurrent stroke within 14 days: Date of Haemorrhagic stroke (yyyy/mm/dd)

DRSISC

Recurrent stroke within 14 days: Ischaemic recurrent stroke

DRSISCD

Recurrent stroke within 14 days: Date of Ischaemic recurrent stroke (yyyy/mm/dd)

DRSUNK

Recurrent stroke within 14 days: Unknown type

DRSUNKD

Recurrent stroke within 14 days: Date of Unknown type (yyyy/mm/dd)

DSCH

Data collected on 14 day/discharge form about treatments given in hospital: Non trial subcutaneous heparin (Y/N)

DSIDE

Data collected on 14 day/discharge form about treatments given in hospital: Other side effect (Y/N)

DSIDED

Data collected on 14 day/discharge form about treatments given in hospital: Date of Other side effect

DSIDEX

Data collected on 14 day/discharge form about treatments given in hospital: Comment of Other side effect

DSTER

Data collected on 14 day/discharge form about treatments given in hospital: Steroids (Y/N)

DTHROMB

Data collected on 14 day/discharge form about treatments given in hospital: Thrombolysis (Y/N)

DVT14

Indicator variables for specific causes of death: Indicator of deep vein thrombosis on discharge form

EXPD14

Other data and derived variables: Predicted probability of death at 14 days

EXPD6

Other data and derived variables: Predicted probability of death at 6 month

EXPDD

Other data and derived variables: Predicted probability of death/dependence at 6 month

FAP

Data collected at 6 months: On antiplatelet drugs

FDEAD

Data collected at 6 months: Dead at six month follow-up (Y/N)

FDEADC

Data collected at 6 months: Cause of death (1-Initial stroke /2-Recurrent stroke (ischaemic or unknown) /3-Recurrent stroke (haemorrhagic) /4-Pneumonia /5-Coronary heart disease /6-Pulmonary embolism /7-Other vascular or unknown /8-Non-vascular /0-unknown)

FDEADD

Data collected at 6 months: Date of death; NOTE: this death is not necessarily within 6 months of randomisation

FDEADX

Data collected at 6 months: Comment on death

FDENNIS

Data collected at 6 months: Dependent at 6 month follow-up (Y/N)

FLASTD

Data collected at 6 months: Date of last contact

FOAC

Data collected at 6 months: On anticoagulants

FPLACE

Data collected at 6 months: Place of residance at 6 month follow-up ( A-Home /B-Relatives home /C-Residential care /D-Nursing home /E-Other hospital departments /U-Unknown)

FRECOVER

Data collected at 6 months: Fully recovered at 6 month follow-up (Y/N)

FU1_COMP

Other data and derived variables: Date discharge form completed

FU1_RECD

Other data and derived variables: Date discharge form received

FU2_DONE

Other data and derived variables: Date 6 month follow-up done

H14

Indicator variables for specific causes of death: Cerebral bleed/heamorrhagic stroke within 14 days; this is slightly wider definition than DRSH an is used for analysis of cerebral bleeds

HOSPNUM

Randomisation data: Hospital number

HOURLOCAL

Randomisation data: Local time – hours

HTI14

Indicator variables for specific causes of death: Indicator of haemorrhagic transformation within 14 days

ID14

Other data and derived variables: Indicator of death at 14 days

ISC14

Indicator variables for specific causes of death: Indicator of ischaemic stroke within 14 days

MINLOCAL

Randomisation data: Local time – minutes

NCB14

Indicator variables for specific causes of death: Indicator of any non-cerebral bleed within 14 days

NCCODE

Other data and derived variables: Coding of compliance (see Table 3) doi:10.1186/1745-6215-13-24

NK14

Indicator variables for specific causes of death: Indicator of indeterminate stroke within 14 days

OCCODE

Other data and derived variables: Six month outcome ( 1-dead /2-dependent /3-not recovered /4-recovered /8 or 9 – missing status

ONDRUG

Data collected on 14 day/discharge form about treatments given in hospital: Estimate of time in days on trial treatment

PE14

Indicator variables for specific causes of death: Indicator of pulmonary embolism within 14 days

RASP3

Randomisation data: Aspirin within 3 days prior to randomisation (Y/N)

RATRIAL

Randomisation data: Atrial fibrillation (Y/N); not coded for pilot phase - 984 patients

RCONSC

Randomisation data: Conscious state at randomisation (F - fully alert, D - drowsy, U - unconscious)

RCT

Randomisation data: CT before randomisation (Y/N)

RDATE

Randomisation data: Date of randomisation

RDEF1

Randomisation data: Face deficit (Y/N/C=can't assess)

RDEF2

Randomisation data: Arm/hand deficit (Y/N/C=can't assess)

RDEF3

Randomisation data: Leg/foot deficit (Y/N/C=can't assess)

RDEF4

Randomisation data: Dysphasia (Y/N/C=can't assess)

RDEF5

Randomisation data: Hemianopia (Y/N/C=can't assess)

RDEF6

Randomisation data: Visuospatial disorder (Y/N/C=can't assess)

RDEF7

Randomisation data: Brainstem/cerebellar signs (Y/N/C=can't assess)

RDEF8

Randomisation data: Other deficit (Y/N/C=can't assess)

RDELAY

Randomisation data: Delay between stroke and randomisation in hours

RHEP24

Randomisation data: Heparin within 24 hours prior to randomisation (Y/N)

RSBP

Randomisation data: Systolic blood pressure at randomisation (mmHg)

RSLEEP

Randomisation data: Symptoms noted on waking (Y/N)

RVISINF

Randomisation data: Infarct visible on CT (Y/N)

RXASP

Randomisation data: Trial aspirin allocated (Y/N)

RXHEP

Randomisation data: Trial heparin allocated (M/L/N) \[M is coded as H=high in pilot\]

SET14D

Other data and derived variables: Know to be dead or alive at 14 days (1=Yes, 0=No); this does not necessarily mean that we know outcome at 6 monts – see OCCODE for this

SEX

Randomisation data: M=male; F=female

STRK14

Indicator variables for specific causes of death: Indicator of any stroke within 14 days

STYPE

Randomisation data: Stroke subtype (TACS/PACS/POCS/LACS/other)

TD

Other data and derived variables: Time of death or censoring in days

TRAN14

Indicator variables for specific causes of death: Indicator of major non-cerebral bleed within 14 days

...

Details

Obtained from Sandercock, Peter; Niewada, Maciej; Czlonkowska, Anna. (2011). International Stroke Trial database (version 2), [dataset]. University of Edinburgh. Department of Clinical Neurosciences. doi:10.7488/ds/104 Under ODC-by licence

Author(s)

Sandercock P et al. [email protected]

References

doi:10.7488/ds/104

---

print.RESIDE

Description

S3 override for print RESIDE

Usage

## S3 method for class 'RESIDE'
print(x, ...)

Arguments

x	an object of class RESIDE
...	Other parameters, full = TRUE prints every category of the categorical variables, Default: FALSE

Details

S3 Override for RESIDE Class, prints the overall summary followed by the marginal distributions of each data frame. By default categorical variables with more than 10 categories only print the 10 most common categories, use full = TRUE to print every category.

Value

The RESIDE object, invisibly. Called to print to the terminal.

Examples

print(
  marginal_distributions <- get_marginal_distributions(
    IST,
    variables = c(
      "SEX",
      "AGE",
      "ID14",
      "RSBP",
      "RATRIAL"
    )
  )
)
print(marginal_distributions, full = TRUE)

---

print.summary.RESIDE

Description

S3 override for print summary.RESIDE

Usage

## S3 method for class 'summary.RESIDE'
print(x, ...)

Arguments

x	an object of class summary.RESIDE
...	Other parameters currently none are used

Details

S3 Override for summary.RESIDE Class, prints the overall summary followed by a table summarising each data frame.

Value

The summary.RESIDE object, invisibly. Called to print to the terminal.

See Also

summary.RESIDE

---

summary.RESIDE

Description

S3 override for summary RESIDE

Usage

## S3 method for class 'RESIDE'
summary(object, ...)

Arguments

object	an object of class RESIDE
...	Other parameters currently none are used

Details

S3 Override for RESIDE Class, a higher level summary than print.RESIDE. For each data frame it gives the number of rows, subjects and variables, the number of each type of variable, the number of date variables and the number of variables with missing data.

Value

An object of class summary.RESIDE, a list containing overall, a data frame of the overall summary, and data_frames, a data frame with a row for each data frame.

See Also

print.RESIDE

Examples

summary(
  get_marginal_distributions(
    IST,
    variables = c(
      "SEX",
      "AGE",
      "ID14",
      "RSBP",
      "RATRIAL"
    )
  )
)

---

Synthesise data from marginal distributions

Description

Allows the synthesis of data from marginal distributions obtained from a Trusted Research Environment (TRE)

Usage

synthesise_data(marginals, correlation_matrix = NULL, correlations = NULL, ...)

synthesize_data(marginals, correlation_matrix = NULL, correlations = NULL, ...)

Arguments

marginals	an object of class RESIDE
correlation_matrix	No longer supported, use correlations. Default: NULL
correlations	A list of correlations created with the correlation function, Default: NULL
...	Additional parameters currently none are used.

Details

This function will synthesise a dataset from marginals imported using import_marginal_distributions. By default the dataset will not contain correlations, however user specified correlations can be added using the correlations parameter, see correlation. Categorical variables are correlated using a single category, specified with factor_name.x or factor_name.y. Correlated variables are synthesised together, one row per subject, and joined to each data frame by subject. Correlated variables therefore take a single value per subject within each data frame. It is not possible to entirely maintain the marginal distributions when specifying correlations.

Value

a data frame of simulated data, or a named list of data frames for marginals from multiple data frames.

See Also

correlation

Examples

marginals <- get_marginal_distributions(
  IST,
  variables = c("SEX", "AGE", "RSBP", "RATRIAL")
)
df <- synthesise_data(marginals)
df_cor <- synthesise_data(
  marginals,
  correlations = list(
    correlation("AGE", "RSBP", 0.3),
    correlation("SEX", "AGE", -0.2, factor_name.x = "M")
  )
)

---