convert_notes: Extracts information from notes free text.

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

Extracts information from notes free text.

Description

Analyzes notes loaded using load_notes or load_lno. Extracts information from the free text present in abc_rep_txt, where abc stands for the three letter abbreviation of the given type of note. An array of string is provided using the anchors argument. The function will return as many columns as there are anchor points. Each column will contain the text between the given anchor point and the next following anchor point. This way the free text report is split into corresponding smaller texts. By default, these are the common standard elements of given note types. Here are provided potential anchor points for the given types of notes:

Cardiology:

c("Report Number:", "Report Status:", "Type:", "Date:", "Ordering Provider:", "SYSTOLIC BLOOD PRESSURE", "DIASTOLIC BLOOD PRESSURE", "VENTRICULAR RATE EKG/MIN", "ATRIAL RATE", "PR INTERVAL", "QRS DURATION", "QT INTERVAL", "QTC INTERVAL", "P AXIS", "R AXIS", "T WAVE AXIS", "LOC", "DX:", "REF:", "Electronically Signed", "report_end")

Discharge:

c("***This text report", "Patient Information", "Physician Discharge Summary", "Surgeries this Admission", "Items for Post-Hospitalization Follow-Up:", "Pending Results", "Hospital Course", "ED Course:", "Diagnosis", "Prescriptions prior to admission", "Family History:", "Physical Exam on Admission:", "Discharge Exam", "report_end")

Endoscopy:

c("NAME:", "DATE:", "Patient Information", "report_end")

History & Physical:

c("***This text report", "Patient Information", "H&P by", "Author:", "Service:", "Author Type:", "Filed:", "Note Time:", "Status:", "Editor:", "report_end")

Operative:

c("NAME:", "UNIT NO:, "DATE:", "SURGEON:", "ASST:", "PREOPERATIVE DIAGNOSIS:", "POSTOPERATIVE DIAGNOSIS:", "NAME OF OPERATION:", "ANESTHESIA:", "INDICATIONS", "OPERATIVE FINDINGS:", "DESCRIPTION OF PROCEDURE:", "Electronically Signed", "report_end")

Pathology:

c("Accession Number:", "Report Status:", "Type:", "Report:", "CASE:", "PATIENT:", "Date", "Source Care Unit:", "Path Subspecialty Service:", "Results To:", "Signed Out by:", "CLINICAL DATA:", "FINAL DIAGNOSIS:", "GROSS DESCRIPTION:", "report_end")

Progress:

c("***This text report", "Patient Information", "History", "Overview", "Progress Notes", "Medications", "Relevant Orders", "Level of Service", "report_end")

Pulmonary:

c("The Pulmonary document", "Name:", "Unit #:", "Date:", "Location:", "Smoking Status:", "Pack Years:", "SPIROMETRY:", "LUNG VOLUMES:", "DIFFUSION:", "PLETHYSMOGRAPHY:" "Pulmonary Function Test Interpretation", "Spirometry", "report_end")

Radiology:

c("Exam Code", "Ordering Provider", "HISTORY", "Associated Reports", "Report Below", "REASON", "REPORT", "TECHNIQUE", "COMPARISON", "FINDINGS", "IMPRESSION", "RECOMMENDATION", "SIGNATURES", "report_end")

Visit:

c("***This text report", "Reason for Visit", "Reason for Visit", "Vital Signs", "Chief Complaint", "History", "Overview", "Medications", "Relevant Orders", "Level of Service", "report_end"

LMR:

c("Subject", "Patient Name:", "Reason for visit", "report_end"

However, these may be modified and extended to include sections of interest, i.e. if a given score is reported in a standard fashion, then adding this phrase (i.e. "CAD-RADS") would create a column where the text following this statement is returned. After this the resulting columns can be easily cleaned up if needed. Be aware to always include "report_end" in the anchors array, to provide the function of the last occurring statement in the report.

Usage

convert_notes(
  d,
  code = NULL,
  anchors = NULL,
  nThread = parallel::detectCores() - 1
)

Arguments

d	data.table, database containing notes loaded using the load_notes function.
code	string vector, column name containing the results, which should be "abc_rep_txt", where abc stands for the three letter abbreviation of the given type of note.
anchors	string array, elements to search for in the text report.
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.

Value

data.table, with new columns corresponding to elements in anchors.

Examples

## Not run: 
#Create columns with specific parts of the radiological report defined by anchors
data_rad_parsed <- convert_notes(d = data_rad, code = "rad_rep_txt",
anchors = c("Exam Code", "Ordering Provider", "HISTORY", "Associated Reports",
"Report Below", "REASON", "REPORT", "TECHNIQUE", "COMPARISON", "FINDINGS",
"IMPRESSION", "RECOMMENDATION", "SIGNATURES", "report_end"), nThread = 2)

## End(Not run)

parseRPDR documentation built on June 24, 2024, 5:16 p.m.