JMIR Publications

ABSTRACT

Background:

Opioid overdose has become a global public health emergency, with the United States experiencing particularly high rates of morbidity and mortality due to both prescription and illicit opioid use. Traditional public health monitoring systems often fail to provide real-time insights, limiting their capacity for early detection and intervention. Social media platforms, especially Reddit, offer a promising alternative for timely toxicovigilance due to the abundance of user-generated, real-time content.

Objective:

This study aims to explore the use of Reddit as a real-time, high-volume source for toxicovigilance and to develop an automated system that can classify and analyze opioid-related social media posts to detect behavioral patterns and monitor the evolution of public discourse on opioid use.

Methods:

To investigate the evolving social media chatter discourse around opioid use, we collected a large-scale dataset from Reddit spanning six years, from January 1, 2018, to December 30, 2023. Using a comprehensive opioid lexicon—including formal drug names, street slang, common misspellings, and abbreviations—we filtered relevant chatters post for further analysis. A subset of this data was manually annotated according to well-defined annotation guidelines into four distinct categories: Self-abuse (chatter describing his/her own experience with opioid use or overdose), External-abuse (use by someone close, such as a friend or family member), Information (general or factual knowledge about opioids), and Unrelated (content not contextually relevant to opioid use). The distribution across categories was as follows: 37.21% Self-abuse, 27.25% External-abuse, 27.57% Information, and 7.97% Unrelated. To automate the classification of opioid-related chatter, we developed a robust NLP pipeline leveraging classical machine learning algorithms, deep learning models, transformer-based architectures, and fine-tuned a state-of-the-art large language model (OpenAI GPT-3.5 Turbo). In the final stage, the trained LLM was deployed on an unlabeled dataset comprising 74,975 additional Reddit chatter entries. This enabled a detailed temporal analysis of opioid-related discussions over the six-year period, uncovering trends and shifts in public perception, self-reported use, external reported use, and information sharing around opioid drugs. This methodology demonstrates the power of combining manual annotation with cutting-edge language models for real-time toxicovigilance and public health monitoring.

Results:

The fine-tuned GPT-3.5 Turbo model achieved the highest classification accuracy of 0.86, outperforming the mBERT model (0.81) by representing a performance improvement of 6.17% over the Transformer model. The temporal analysis of the unlabeled data revealed evolving trends in opioid-related discussions, indicating shifts in user behavior and overdose-related chatter over time.

Conclusions:

This study demonstrates the potential of integrating advanced NLP techniques and LLMs with social media data to support real-time public health surveillance. Reddit provides a valuable platform for identifying emerging trends in opioid use and overdose risk. The proposed system offers a proactive tool for researchers, clinicians, and policymakers to better understand and respond to the opioid crisis.