Skip to contents

Patient Record Consistency – Output Generation

Source: R/prc_output.R
prc_output.Rd

Using the tabular output generated by prc_process, this function will build a graph to visualize the results. Each function configuration will output a bespoke ggplot. Theming can be adjusted by the user after the graph has been output using + theme(). Most graphs can also be made interactive using make_interactive_squba()

Usage

prc_output(
  process_output,
  dist_from_stat = "mean",
  event_filter = NULL,
  large_n = FALSE,
  large_n_sites = NULL
)

Arguments

process_output

tabular input || required

The tabular output produced by pes_process

Note any patient-level results generated are not intended to be used with this function.

dist_from_stat

string || defaults to mean

A string indicating the statistic from which distance should be measured for the Multi-Site, Exploratory, Longitudinal check

Acceptable values are mean or median

event_filter

string || defaults to NULL

A string indicating the event co-occurrence type of interest for the analysis. This parameter is required for the following checks:

  • Single Site, Anomaly Detection, Longitudinal

  • Multi-Site, Anomaly Detection, Longitudinal

Acceptable values are a, b, both, or neither

large_n

boolean || defaults to FALSE

For Multi-Site analyses, a boolean indicating whether the large N visualization, intended for a high volume of sites, should be used. This visualization will produce high level summaries across all sites, with an option to add specific site comparators via the large_n_sites parameter.

large_n_sites

vector || defaults to NULL

When large_n = TRUE, a vector of site names that can add site-level information to the plot for comparison across the high level summary information

Value

This function will produce a graph to visualize the results from prc_process based on the parameters provided. The default output is typically a static ggplot or gt object, but interactive elements can be activated by passing the plot through make_interactive_squba. For a more detailed description of output specific to each check type, see the PEDSpace metadata repository

Examples


#' Source setup file
source(system.file('setup.R', package = 'patientrecordconsistency'))

#' Create in-memory RSQLite database using data in extdata directory
conn <- mk_testdb_omop()

#' Establish connection to database and generate internal configurations
initialize_dq_session(session_name = 'prc_process_test',
                      working_directory = my_directory,
                      db_conn = conn,
                      is_json = FALSE,
                      file_subdirectory = my_file_folder,
                      cdm_schema = NA)
#> Connected to: :memory:@NA

#' Build mock study cohort
cohort <- cdm_tbl('person') %>% dplyr::distinct(person_id) %>%
  dplyr::mutate(start_date = as.Date(-5000), # RSQLite does not store date objects,
                                      # hence the numerics
                end_date = as.Date(15000),
                site = ifelse(person_id %in% c(1:6), 'synth1', 'synth2'))

#' Build function input table
prc_events <- tidyr::tibble(event = c('a', 'b'),
                            event_label = c('hypertension', 'inpatient/ED visit'),
                            domain_tbl = c('condition_occurrence', 'visit_occurrence'),
                            concept_field = c('condition_concept_id', 'visit_concept_id'),
                            date_field = c('condition_start_date', 'visit_start_date'),
                            vocabulary_field = c(NA, NA),
                            codeset_name = c('dx_hypertension', 'visit_edip'),
                            filter_logic = c(NA, NA))

#' Execute `prc_process` function
#' This example will use the single site, exploratory, cross sectional
#' configuration
prc_process_example <- prc_process(cohort = cohort,
                                   multi_or_single_site = 'single',
                                   anomaly_or_exploratory = 'exploratory',
                                   time = FALSE,
                                   omop_or_pcornet = 'omop',
                                   prc_event_file = prc_events) %>%
  suppressMessages()
#> ┌ Output Function Details ──────────────────────────────────────┐
#> │ You can optionally use this dataframe in the accompanying     │
#> │ `prc_output` function. Here are the parameters you will need: │
#> │                                                               │
#> │ Always Required: process_output                               │
#> │                                                               │
#> │ See ?prc_output for more details.                             │
#> └───────────────────────────────────────────────────────────────┘

prc_process_example
#> # A tibble: 5 × 8
#>   site     event_a_num event_a_name event_b_num event_b_name     pt_ct total_pts
#>   <chr>          <dbl> <chr>              <dbl> <chr>            <int>     <int>
#> 1 combined           0 hypertension           0 inpatient/ED vi…     6        12
#> 2 combined           0 hypertension           1 inpatient/ED vi…     1        12
#> 3 combined           1 hypertension           1 inpatient/ED vi…     2        12
#> 4 combined           1 hypertension           2 inpatient/ED vi…     2        12
#> 5 combined           1 hypertension           5 inpatient/ED vi…     1        12
#> # ℹ 1 more variable: output_function <chr>

#' Execute `prc_output` function
prc_output_example <- prc_output(process_output = prc_process_example)

prc_output_example


#' Easily convert the graph into an interactive ggiraph or plotly object with
#' `make_interactive_squba()`

make_interactive_squba(prc_output_example)