ATEdrug: A reliable human-in-the-loop annotation scheme for aspect term extraction and polarity detection in drug reviews

Gunjan Ansari, Chandni Saxena, Shahab Saquib Sohail, Zain Hussain, Aziz Sheikh and Amir Hussain

PLOS ONE, 2026, vol. 21, issue 4, 1-16

Abstract: Pharmacovigilance is vital for post-market drug safety monitoring. Traditional trials inadequately capture adverse reactions. Patient-generated opinions offer valuable insights but pose challenges due to limited availability, inefficient annotation scheme, and volume of drug reviews, indicating the need of automation. To address the challenge posed by the limited availability of annotated drug reviews, we introduce a novel mechanism of annotating drug reviews dataset through human intervention, specifically tailored for Aspect Term Extraction (ATE) and Polarity Detection (PD) in three medical conditions: Depression, Arthritis and Birth control, thereby making it publicly available to facilitate future research. We first designed, ATEdrug, an automated annotation scheme with minimal human intervention by employing an expert-driven rule-based approach. We further deploy ATEdrug for sequence labelling and classification tasks in drug reviews. The automatically labeled data is further used for training and testing of deep learning models: BERT, BioBERT, and ClinicalBERT, for clinical use. We manually evaluate the results through annotator agreement to validate the effectiveness of ATEdrug. The labeled dataset is further used to construct transformer-based models. Our proposed model is reliable, safe and trustworthy for healthcare domain, thereby eliminating the compromises with hallucination and data security through generative AI models. All experimental codes and data are available on GitHub (https://github.com/CMOONCS/ateDrug.git).

Date: 2026
References: View complete reference list from CitEc
Citations:

Downloads: (external link)
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0344296 (text/html)
https://journals.plos.org/plosone/article/file?id= ... 44296&type=printable (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:plo:pone00:0344296

DOI: 10.1371/journal.pone.0344296

Access Statistics for this article

More articles in PLOS ONE from Public Library of Science
Bibliographic data for series maintained by plosone ().