Using Transformer-Based Topic Modeling to Examine Discussions of Delta-8 Tetrahydrocannabinol: Content Analysis

Area of Science:

• Cannabinoid research
• Public health surveillance
• Natural language processing applications

Background:

• Delta-8 tetrahydrocannabinol (THC) is a psychoactive cannabinoid found naturally in cannabis and synthetically produced from hemp.
• Hemp-derived delta-8 THC products are widely available, even where delta-9 THC is illegal.
• Health effects and processing concerns of delta-8 THC remain understudied.

Purpose of the Study:

• To implement a novel transformer-based topic modeling approach.
• To identify and describe emerging trends and discussion topics about delta-8 THC from social media.
• To analyze potential symptoms and adverse health outcomes associated with delta-8 THC use.

Main Methods:

• Isolated 41,191 delta-8 THC posts from Reddit forums (2008-2021).
• Utilized unsupervised, state-of-the-art transformer-based models for topic clustering.
• Validated results with human subject matter experts.

Main Results:

• Identified 81 topics, with prevalent themes including product discussions, cannabinoid comparisons, and safety warnings.
• Approximately 5% of posts (n=1220) discussed health symptoms like anxiety, sleep disturbance, and breathing problems.
• Mentions of delta-8 THC on Reddit increased dramatically from 2019 to 2021.

Conclusions:

• Provides insights into emerging public health concerns surrounding delta-8 THC.
• Demonstrates the utility of transformer-based learning for identifying health trends from unstructured data.
• Highlights the potential for improved timeliness in identifying health concerns related to novel substances.