Development of a Medical Incident Report Corpus with Intention and Factuality Annotation

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 4578–4584 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 4578 Development of a Medical Incident Report Corpus with Intention and Factuality Annotation Hongkuan Zhang†, Ryohei Sasano†, Koichi Takeda†, Zoie Shui-Yee Wong‡ † Graduate School of Informatics, Nagoya University, Nagoya, Japan ‡ Graduate School of Public Health, St.Luke’s International University, Tokyo, Japan zhang.hongkuan@a.mbox.nagoya-u.ac.jp Abstract Medical incident reports (MIRs) are documents that record what happened in a medical incident. A typical MIR consists of two sections: a structured categorical part and an unstructured text part. Most texts in MIRs describe what medication was intended to be given and what was actually given, because what happened in an incident is largely due to discrepancies between intended and actual medications. Recognizing the intention of clinicians and the factuality of medication is essential to understand the causes of medical incidents and avoid similar incidents in the future. Therefore, we are developing an MIR corpus with annotation of intention and factuality as well as of medication entities and their relations. In this paper, we present our annotation scheme with respect to the definition of medication entities that we take into account, the method to annotate the relations between entities, and the details of the intention and factuality annotation. We then report the annotated corpus consisting of 349 Japanese medical incident reports. Keywords: medical incident reports, intention and factuality annotation, incident comparative table 1. Introduction Medical incidents have been shown to cause significant harm to patients (Kohn et al., 2000). To prevent similar incidents from happening again, several countries have es- tablished national reporting systems to collect medical in- cident reports (MIRs). Among the gathered incident types, those caused by medication errors are one of the most fre- quently occurring types (Grissinger, 2010). Figure 1 shows an example of a medication-related MIR that contains two sections: a structured categorical information part includ- ing patient age, sex, incident date, type, severity level, etc., and an unstructured text part that records what happened in the incident. While both the structured and unstructured parts contain important information on the incidents, it is relatively difficult to extract information from the unstruc- tured part because it is written in natural language. There- fore, accurate techniques for information extraction from the unstructured part of medication-related incident reports are needed. Recognizing the intention of clinicians and the factuality of medication1 is essential to extract important information from an MIR. For example, the text shown in the bottom half of Figure 1 describes that clinicians intended to use Rebamipide 100mg but Sennoside 12mg were actually de- livered, which indicates that the drug and its amount were wrong. Table 1 summarizes the intention and factuality of this incident. Our goal is to automatically construct this kind of comparative table from incident reports. To achieve this, we think that three different technologies are required: 1) medication entity recognition, 2) entity relation detec- tion, and 3) intention/factuality analysis. Let us …

Content truncated to 500 words