# Assessing patient perceptions of off-label cannabidiol use for insomnia through sentiment analysis

## TL;DR

This study uses sentiment analysis of social media posts to explore how patients perceive the effectiveness of cannabidiol (CBD) for treating insomnia, finding generally positive feedback.

## Contribution

The novel use of sentiment analysis to assess patient perceptions of off-label CBD use for insomnia, complementing limited clinical evidence.

## Key findings

- Sentiment analysis of tweets revealed positive patient perceptions of CBD for insomnia.
- Topic modeling identified themes like perceived efficacy for anxiety, pain, and insomnia.
- The study highlights the potential of sentiment analysis as a low-cost tool for understanding patient experiences.

## Abstract

Recent global policy shifts have greatly expanded access to and the market for medicinal cannabis, broadening its availability for therapeutic use. These changes have led to a notable increase in off-label prescriptions of cannabidiol (CBD) based products, particularly for conditions such as epilepsy, anxiety, depression, insomnia, and chronic pain. Despite its growing popularity, clinical studies investigating the efficacy of CBD for insomnia remain limited, raising questions about its suitability for decision support. In this scenario, sentiment analysis provides an additional, low-cost, and measurable outcome of patients’ perceptions, which has proven valuable in offering an additional layer of understanding regarding the perceived effectiveness of treatments. Therefore, the objective of this study was to assess patients' perception regarding off-label CBD use for insomnia through sentiment analysis.

English-language tweets related to CBD use for insomnia were collected from June 25, 2018, to January 9, 2023. The data was cleaned, and a representative subset of comments was manually labeled by experts. Then, a ROBERTA-based model was trained to automatically classify the remaining tweets in the database as either signals or noise. Finally, topic modeling and sentiment analysis were performed on the database of signal tweets.

From the 74,562 unique comments retrieved, 25,005 were classified as relevant based on both manual labeling and predictions made by the validated model. Topic modeling revealed eleven major themes, including the perceived efficacy of CBD for anxiety, pain, and insomnia, as well as practical considerations regarding treatment routines and preparation preferences. Sentiment analysis revealed positive feedback regarding CBD's use for insomnia, indicating that most users perceive it as an effective treatment for this condition.

This study offers valuable insights into patient perceptions of off-label CBD use for insomnia, suggesting that CBD may indeed be beneficial for this condition, which aligns with existing, albeit limited, experimental evidence. Consequently, further research is necessary to confirm its efficacy. In this scenario, the application of sentiment analysis emerges as an effective tool for assessing patient perceptions, providing a richer context that complements the still limited evidence from traditional clinical trials on CBD use for insomnia.

Trial registration.

Not applicable.

The online version contains supplementary material available at 10.1186/s42238-025-00306-7.