Skip to main content

Table 7 Annotation times (seconds per text) on 100 new trial texts and 100 clinical cases

From: Hybrid natural language processing tool for semantic annotation of medical texts in Spanish

Text

Average

Average

RoBERTa

EriBERTa

CLIN-X-ES

mDeBERTa

type

bytes/text

tokens/text

EHR

  

vs 3

EudraCT

2074.92

298.51

24”

24”

96”

153”

 

(± 763.67)

(± 115.71)

    

Clinical cases

1945.18

278.31

28”

29”

111”

129”

 

(± 593.39)

(± 80.55)

    
  1. For each model architecture, we timed the annotation process in which six models were applied (for UMLS entities, medication information, temporal entities, negation/speculation, miscellaneous entities and experiencer/event temporality attributes)
Back to article page

BMC Bioinformatics

ISSN: 1471-2105

Contact us