find_exam: Find exam data within a given timeframe using parallel CPU...

In parseRPDR: Parse and Manipulate Research Patient Data Registry ('RPDR') Text Queries

Find exam data within a given timeframe using parallel CPU computing and possibly shared RAM management.

Description

Finds all, earliest or closest examination to a given timepoints using parallel computing

Usage

find_exam(
  d_from,
  d_to,
  d_from_ID = "ID_MERGE",
  d_to_ID = "ID_MERGE",
  d_from_time = "time_rad_exam",
  d_to_time = "time_enc_admit",
  time_diff_name = "timediff_exam_to_db",
  before = TRUE,
  after = TRUE,
  time = 1,
  time_unit = "days",
  multiple = "closest",
  add_column = NULL,
  keep_data = FALSE,
  nThread = parallel::detectCores() - 1,
  shared_RAM = FALSE
)

Arguments

d_from	data table, the database which is searched to find examinations within the timeframe.
d_to	data table, the database to which we wish to find examinations within the timeframe.
d_from_ID	string, column name of the patient ID column in d_from. Defaults to ID_MERGE.
d_to_ID	string, column name of the patient ID column in d_to. Defaults to ID_MERGE.
d_from_time	string, column name of the time variable column in d_from. Defaults to time_rad_exam.
d_to_time	string, column name of the time variable column in d_to. Defaults to time_enc_admit.
time_diff_name	string, column name of the new column created which holds the time difference between the exam and the time provided by d_to. Defaults to timediff_exam_to_db.
before	boolean, should times before the given time be considered. Defaults to TRUE.
after	boolean, should times after the given time be considered. Defaults to TRUE.
time	integer, the timeframe considered between the exam and the d_to timepoints. Defaults to 1.
time_unit	string, the unit of time used. Time variables are in d_to and d_from are truncated to the supplied time unit. For example: "2005-09-18 08:15:01 PDT" would be truncated to "2005-09-18 PDT" if time_unit is set to days. Then the time differences is calculated using difftime passing the argument to units. The following time units are supported: "secs", "mins", "hours", "days", "months" and "years" are supported. Defautls to days.
multiple	string, which exams to give back. closest gives back the exam closest to the time provided by d_to. all gives back all occurrences within the timeframe. earliest the earliest exam within the timeframe. In case of ties for closest or earliest, all are returned. Defaults to closest.
add_column	string, a column name in d_to to add to the output. Defaults to NULL.
keep_data	boolean, whether to include empty rows with only the d_from_ID column filed out for cases that have data in the d_from, but not within the time range. Defaults to FALSE.
nThread	integer, number of threads to use for parallelization. If it is set to 1, then no parallel backends are created and the function is executed sequentially.
shared_RAM	boolean, whether to use shared memory during parallelization using the bigmemory package. This allows to process d_from and/or d_to datasets with >1M rows. Be aware that shared RAM usually results in slower run times, therefore by default it is set to FALSE, but it allows to run more threads when the datasets are large providing overall faster run times. Be aware that the optimal number of clusters might be different setting it to TRUE or FALSE, and this has to be determined empirically per machine. The feature is very unstable and therefore should only be tried if there is no other option

Value

data table, with d_from filtered to ones only within the timeframe. The columns of d_from are returned with the corresponding time column in data_to where the rows are instances which comply with the time constraints specified by the function. An additional column specified in time_diff_name is also returned, which shows the time difference between the time column in d_from and d_to for that given case. Also the time column from d_to specified by d_to_time is returned under the name of time_to_db. An additional column specified in add_column may be added from data_to to the data table.

Examples

## Not run: 
#Filter encounters for first emergency visits at one of MGH's ED departments
data_enc_ED <- data_enc[enc_clinic == "MGH EMERGENCY 10020010608"]
data_enc_ED <- data_enc_ED[!duplicated(data_enc_ED$ID_MERGE)]

#Find all radiological examinations within 3 day of the ED registration
rdt_ED <- find_exam(d_from = data_rdt, d_to = data_enc_ED,
d_from_ID = "ID_MERGE", d_to_ID = "ID_MERGE",
d_from_time = "time_rdt_exam", d_to_time = "time_enc_admit", time_diff_name = "time_diff_ED_rdt",
before = TRUE, after = TRUE, time = 3, time_unit = "days", multiple = "all",
nThread = 2, shared_RAM = FALSE)

#Find earliest radiological examinations within 3 day of the ED registration
rdt_ED <- find_exam(d_from = data_rdt, d_to = data_enc_ED,
d_from_ID = "ID_MERGE", d_to_ID = "ID_MERGE",
d_from_time = "time_rdt_exam", d_to_time = "time_enc_admit", time_diff_name = "time_diff_ED_rdt",
before = TRUE, after = TRUE, time = 3, time_unit = "days", multiple = "earliest",
nThread = 2, shared_RAM = FALSE)

#Find closest radiological examinations on or after 1 day of the ED registration
#and add primary diagnosis column from encounters
rdt_ED <- find_exam(d_from = data_rdt, d_to = data_enc_ED,
d_from_ID = "ID_MERGE", d_to_ID = "ID_MERGE",
d_from_time = "time_rdt_exam", d_to_time = "time_enc_admit", time_diff_name = "time_diff_ED_rdt",
before = FALSE, after = TRUE, time = 1, time_unit = "days", multiple = "earliest",
add_column = "enc_diag_princ", nThread = 2, shared_RAM = FALSE)

#Find closest radiological examinations on or after 1 day of the ED registration
#but also provide empty rows for patients with exam data but not within the timeframe
rdt_ED <- find_exam(d_from = data_rdt, d_to = data_enc_ED,
d_from_ID = "ID_MERGE", d_to_ID = "ID_MERGE",
d_from_time = "time_rdt_exam", d_to_time = "time_enc_admit", time_diff_name = "time_diff_ED_rdt",
before = FALSE, after = TRUE, time = 1, time_unit = "days", multiple = "earliest",
add_column = "enc_diag_princ", keep_data = TRUE nThread = 2, shared_RAM = FALSE)

## End(Not run)

parseRPDR documentation built on March 31, 2023, 11:36 p.m.