JMIR Publications

ABSTRACT

Background:

Licensed drugs may potentially cause unexpected adverse reactions to human patients resulting in morbidity, risk of mortality, therapy disruptions and prolonged hospital stay. Officially approved drug package inserts provide the identified adverse reactions obtained from randomized controlled clinical trials with high evidence levels and worldwide post-marketing experience. Formal representation of the adverse drug reactions (ADRs) enclosed in semi-structured package inserts will enable deep recognition of the side effects and rational drug use, substantially reduce morbidity and decrease societal costs.

Objective:

This paper aims to present an ontological organization of traceable ADRs information extracted from licensed package inserts, as well as provide machine-understandable knowledge for bioinformatics analysis, semantic retrieval and intelligent clinical applications.

Methods:

Based on the essential content of package insert, a generic ADRs ontology model is proposed from 2 dimensions (and 9 sub-dimensions) covering the ADRs information and medication instruction, followed by a customized Python NLP programmed method to retrieve the relevant information enclosed in package inserts. After the biocuration and identification of retrieved data from package insert, an ADRs ontology is automatically built for further bioinformatic analysis.

Results:

We collected 165 package inserts of particular quinolones drugs from the National Medical Products Administration and other drug database in China, and built a specialized ADRs ontology containing 2879 classes and 15711 semantic relations. For each quinolones drug, the reported ADRs information and medication instruction have been logically represented and formally organized in ADRs ontology. To demonstrate its usage, the obtained knowledge source data were further bioinformatically analyzed. Especially, the number and percentage of drug-ADR triples and major ADRs were counted according to every active ingredient of drugs. The top 10 ADRs most frequently observed among quinolones were identified and categorized dependent on the 18 categories defined in the proposal. The occurrence frequency, severity and mitigation method of ADRs explicitly stated in package inserts were also analyzed, as well as the top 5 specific populations pertaining to contraindication for quinolones drugs.

Conclusions:

The ontological representation and organization using officially approved information from the drug package inserts enables the identification and bioinformatic analysis of adverse reactions caused by a specific drug in regard to predefined ADRs ontology classes and semantic relations. The resulting ontology-based ADRs knowledge source classifies drug-specific adverse reactions, supports better ADRs understanding and safe medication in an automated and intelligent way.