Testing the feasibility and acceptability of social media and digital therapeutics to decrease vaping behaviors

Project Summary
Use of vaping products (e.g., electronic nicotine delivery systems, e-cigarettes) has been increasing rapidly,
particularly among teens and young adults. With limited information on the long-term effects of vaping
products, health information about vaping has been somewhat unclear in regards to associated health risks.
Teens and young adults may be reluctant to disclose their use of vaping products to parents or health providers
and instead turn to social media to share and seek out information regarding vaping risks and cessation
supports. Given the ubiquitous use of social media platforms among this population and the ability for
advanced artificial intelligence (AI) and natural language processing (NLP) technologies to analyze content
shared on social media platforms, there is strong potential for this information to be leveraged and used to
detect and reach out to those most at risk for negative health outcomes caused by vaping. Thus, our current
proposal outlines the use of detection models to identify teens and young adults socially networking about
vaping, the use of a chatbot to screen for the needs of eligible users, and the use of a digital intervention system
(i.e., quitSTART with an embodied chatbot) aimed to support vaping cessation efforts by increasing risk
awareness and decreasing pro-vaping attitudes. In Aim #1a, we will develop an intelligent detection system by
leveraging state-of-the-art machine learning, deep learning, and NLP techniques for mining massive social
media data on vaping with clinical inputs. This detection system will implement multiple functionalities on
both Twitter and Reddit social media platforms to identify posts regarding the use of vaping products, negative
health outcomes experienced, and interest in vaping cessation. To evaluate the validity and specificity of the
detection model developed on both platforms, we will also conduct surveys among a subsample of those
identified (N=100) to rule out false positives and to gather data on vaping behaviors, social media content
generation about vaping, motivations for vaping product use, and interest in vaping cessation to refine the
developed models in Aim #1b. In Aim #2, we will develop both a chatbot to screen individuals identified on
social media as well as an in-app chatbot to guide users to tailored content, conduct daily assessments and
check-ins, motivate and encourage their cessation efforts, and promote sustained user engagement within a
widely-used evidence-based mobile application (app) intervention for combustible smoking, quitSTART. We
will conduct usability and acceptability testing on both levels of the chatbot among a sample of participants
(N=30) recruited in Aim #1b. In Aim #3, we aim to integrate the developed detection model, chatbot screener,
and adapted mobile app into a streamlined outreach and intervention system, and conduct a randomized
controlled trial (N= 189) to evaluate user engagement with and preliminary efficacy of the digital intervention
on vaping behaviors among teens and young adults. This integrated system has the potential to improve public
health outcomes related to vaping and to inform the feasibility of such chatbot tools to sustain mHealth
intervention engagement.

Grant Number: 5R34DA054725-03
NIH Institute/Center: NIH
Principal Investigator: Patricia Cavazos-Rehg