Understanding Cannabis Mobile Apps: Content, Quality, and Trend (Preprint)

BACKGROUND As cannabis legalization expands across U.S. states, mobile applications have emerged as a key channel for users seeking cannabis-related information, support, and purchasing options. While these apps serve both recreational and medical users, concerns remain about the quality, accuracy, and trustworthiness of the content they deliver—particularly regarding health information. OBJECTIVE This study investigates the landscape of cannabis-related mobile apps by analyzing their content, user perceptions, and categorization. It aims to (1) identify common app types and functions, (2) assess content themes and quality, and (3) understand user experiences and concerns reflected in app reviews. METHODS Using web scraping tools, we compiled metadata for over 4,000 apps from the Apple App Store and Google Play Store. A machine learning pipeline incorporating keyword frequency, large language model (LLM) scoring, and clustering was used to identify cannabis-relevant apps. From this, we selected the top 15 most popular apps from each platform for in-depth analysis. A mixed-methods approach was employed, including inductive qualitative coding of app features and natural language processing (BERTopic) on 15,000+ user reviews to uncover user sentiment and topical trends. RESULTS Cannabis apps clustered into three major categories: consumer-oriented (e.g., dispensary locators, product information), cultivation-oriented (e.g., growing guides, plant recognition), and recovery-oriented (e.g., quit-tracking, peer support). Most apps lacked transparent sourcing of health content, raising concerns about misinformation. Journaling features were found in both consumer and recovery apps, supporting self-monitoring. Topic modeling revealed user appreciation for accessibility, strain details, and dispensary guidance, while negative feedback centered on misinformation, navigation issues, and inconsistent features. CONCLUSIONS Cannabis mobile apps play a dual role as commercial tools and health information providers. However, the lack of validated health content and usability inconsistencies limits their potential as reliable health resources. This study highlights the need for greater transparency, medical oversight, and user-centered design in cannabis-related mHealth tools. Future work should include systematic quality evaluation and engagement with underrepresented or less popular apps used in research settings.