patient_processor_par: Process All Documents for One Patient

In simon-hans/CEDARS: Simple and Efficient Pipeline for Electronic Health Record Annotation

Process All Documents for One Patient

Description

Performs NLP annotations on all documents using previously established cluster, including NegEx and UMLS CUI tags.

Usage

patient_processor_par(
  select_cores,
  cl,
  sub_corpus,
  text_format,
  nlp_engine,
  negex_simp,
  umls_selected,
  max_n_grams_length,
  negex_depth,
  single_core_model
)

Arguments

select_cores	Desired number of cores.
cl	Computing cluster.
sub_corpus	Data frame of text to annotate.
text_format	Text format.
nlp_engine	NLP engine, UDPipe only for now.
negex_simp	Simplifed negex.
umls_selected	Processed UMLS table.
max_n_grams_length	Maximum length of tokens for matching with UMLS concept unique identifiers (CUI's). Shorter values will result in faster processing. If ) is chosen, UMLS CUI tags will not be provided.
negex_depth	Maximum distance between negation item and token to negate. Shorter distances will result in decreased sensitivity but increased specificity for negation.
single_core_model	NLP model in case parallel processing is not used.

Value

NLP annotations dataframe.

simon-hans/CEDARS documentation built on Feb. 14, 2024, 3:16 a.m.