A Hybrid BioBERT-BiLSTM Model for Sentiment Analysis of Malaria Drug Reviews in Pharmacovigilance

Authors

  • Adedeji O. ADEBARE

    Author

  • Olayinka O. OLUSANYA

    Author

  • Ademola A. OMILABU

    Author

  • Iyinoluwa T. IDOWU

    Author

  • Jímọ̀h À. HASSAN

    Author

  • Peter A. IDOWU

    Author

DOI:
https://doi.org/10.33003/qzd2fm11
Abstract
Malaria remains a significant public health challenge in Nigeria, with widespread use of antimalarial drugs leading to frequent adverse drug reactions (ADRs). Current pharmacovigilance systems, such as the Pharmacovigilance Rapid Alert System for Consumer Reporting (PRASCOR), rely on SMS-based reporting and lack comprehensive data collection and analysis capabilities. This study develops a hybrid sentiment analysis model combining BioBERT (Bidirectional Encoder Representations from Transformers for Biomedical Text) and Bidirectional Long Short-Term Memory (BiLSTM) with an attention mechanism to classify malaria drug reviews for enhanced pharmacovigilance. A dataset of 3,194 malaria drug reviews was collected from patients in South-West Nigeria and preprocessed using natural language processing techniques. The proposed BioBERT-BiLSTM model leverages domain-specific contextual embeddings from BioBERT and captures sequential dependencies through BiLSTM, while the attention mechanism focuses on the most informative words for sentiment classification. The model was implemented in Python and evaluated using an 80-20 train-test split, achieving an accuracy of 60.90%, with balanced precision (0.54), recall (0.54), and F1-score (0.54). Comparative analysis against BiLSTM, Random Forest, Support Vector Machine, and Logistic Regression demonstrated the superior performance of the hybrid model across all metrics. The findings show that integrating domain-specific transfer learning with sequential deep learning architectures provides a scalable, accessible, and effective tool for automated pharmacovigilance, improving drug safety monitoring in resource-limited settings.
References
Cover Image
Downloads
Published
25-09-2026
Issue
Vol. 2 No. 2 (2026): December 2026
Section
Articles
License

Copyright (c) 2026 Adedeji O. ADEBARE, Olayinka O. OLUSANYA, Ademola A. OMILABU, Iyinoluwa T. IDOWU, Jímọ̀h À. HASSAN, Peter A. IDOWU (Author)

Creative Commons License

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

How to Cite

[1]
A. O. ADEBARE, O. O. OLUSANYA, A. A. OMILABU, I. T. IDOWU, J. À. HASSAN, and P. A. IDOWU, “A Hybrid BioBERT-BiLSTM Model for Sentiment Analysis of Malaria Drug Reviews in Pharmacovigilance”, FJET, vol. 2, no. 2, pp. 663–670, Sep. 2026, doi: 10.33003/qzd2fm11.