Automated Risk Assessment of Opioid Use: Analysis Using Pre-Trained Transformers on Social Media Data
This paper demonstrates the potential of leveraging social media data and advanced natural language processing models to build reliable systems for opioid overdose risk detection. The BioBERT model with custom attention shows state-of-the-art performance and robustness, offering a powerful tool to support timely intervention and harm reduction strategies in the ongoing opioid crisis.
Authors
- Ildar Batyrshin (ORCID: https://orcid.org/0000-0003-0241-7902)
- Rita Orji (ORCID: https://orcid.org/0000-0001-6152-8034)
- Nailya Kubysheva (ORCID: https://orcid.org/0000-0002-5582-5814)
- Maaz Amjad
- Grigori Sidorov
- Muhammad Ahmad
- Abubakar Siddique
Institutions
- Texas Tech University (US)
- Dalhousie University (CA)
- Kazan Federal University (RU)
- Auckland University of Technology (NZ)
- Instituto Politécnico Nacional (MX)
Publication Details
- Journal
- JMIR Infodemiology
- Published
- 2026-02-19
- DOI
- https://doi.org/10.2196/77783
- Primary Topic
- Opioid Use Disorder Treatment
- Type
- article
- Field-Weighted Citation Impact
- 0.00