Table 2 Definitions and examples of the error categories used for error analysis. Negation terms are underlined; the annotated medical terms are in brackets. Examples translated from the Dutch source text
From: Negation detection in Dutch clinical texts: an evaluation of rule-based and machine learning methods
Category | Definition and example |
|---|---|
Uncommon negation | Negation term rarely occurs in the data ... which \({\underline{\textit{argues against}}}\) a [diagnosis] of RA |
Mminus | In clinical notes, a minus directly following a term indicates negation pale-, [nauseous] -, clammy+ |
Scope | Several medical terms follow/precede a negation, but the negation does not apply to all of them. This often occurs in a list, or simply when the negation occurs much later/earlier in the sentence \({\underline{\textit{no}}}\) abdominal pain, [abrasion on leg] |
Punctuation | Punctuation that is likely to hinder sentence splitting and/or correct recognition of the (scope of the) negation term \({\underline{\textit{no evid. for}}}\) [aneurysm] |
Negation of different term | The negation applies to another term close to the medical term \({\underline{\textit{no}}}\) further investigation of [weight loss] |
Wrong modality | The context is not negation, but hypothetical, historical, or otherwise indication: to \({\underline{\textit{rule out}}}\) [osteopenia] |
Speculation | The clinician expressed uncertainty instead of outright negation \({\underline{\textit{no}}}\) [eczema] after all? |
Ambiguity | The grammatical structure makes it unclear whether a term is negated or not ... \({\underline{\textit{without}}}\) loss of function and [concussion] following... |
Other | The type of error does not fit into any of the other categories |
Annotation error | The original annotator assigned the wrong label |