Optimizing Chatbot Reminder Timing in a Chatbot Cessation Service: A Micro-Randomized Trial

Overview

The goal of this study is to learn whether the timing of weekly chatbot reminders affects user engagement with the chatbot cessation service. Participants enrolled in the routine chatbot cessation service will be included in a micro-randomized trial. The delivery timing of weekly reminders will be randomly assigned across different day types and times of day during the 12-week chatbot cessation service period. The study will use routine chatbot system records to evaluate engagement after reminder delivery, including whether participants respond to the chatbot and how quickly they re-engage with the service.

Sustained user engagement remains a challenge in digital smoking cessation interventions. In this chatbot cessation service, routine weekly AI-generated reminders are used to encourage users to re-engage with chatbot-based smoking cessation support during the 12-week service period. However, the optimal timing for delivering these reminders remains uncertain. This study is an embedded micro-randomized trial conducted within the existing chatbot cessation service. The aim of this study is to evaluate whether the timing of weekly chatbot reminders influences short-term user engagement and to identify optimal reminder delivery schedules. Participants enrolled in the routine chatbot cessation service will be included in the micro-randomized trial through an opt-out approach. The trial will not introduce any additional intervention components, measurements, or study visits beyond the existing chatbot cessation service and follow-up assessments. Participation in the micro-randomized trial will only affect the timing of routine weekly reminders. During the 12-week chatbot cessation service period, excluding the initial welcoming message, the delivery timing of 11 weekly reminders will be randomized across predefined day-type and time-of-day conditions. Reminder strategy will remain unchanged. The primary proximal outcome is meaningful engagement with the chatbot within 24 hours after reminder delivery, as recorded by the chatbot system. Secondary outcomes include reply status, response latency, and meaningful engagement within 48 and 72 hours after reminder delivery. Meaningful engagement is defined as user messages relevant to smoking cessaation. The findings from this study will provide evidence to optimize the timing of routine chatbot reminders and inform future development of optimized reminder strategies and adaptive digital smoking cessation interventions.