Detecting Receptivity for mHealth Interventions
Journal
GetMobile: Mobile Computing and Communications
ISSN
2375-0529
2375-0537
Type
journal article
Date Issued
2023
Author(s)
Abstract
Just-In-Time Adaptive Interventions (JITAI) have the potential to provide effective support for health behavior by delivering the right type and amount of intervention at the right time. The timing of interventions is crucial to ensure that users are receptive and able to use the support provided. Previous research has explored the association of context and user-specific traits on receptivity and built machine-learning models to detect receptivity after the study was completed. However, for effective intervention delivery, JITAI systems need to make in-the-moment decisions about a user's receptivity. In this study, we deployed machinelearning models in a chatbot-based digital coach to predict receptivity for physical-activity interventions. We included a static model that was built before the study and an adaptive model that continuously updated itself during the study. Compared to a control model that sent intervention messages randomly, the machine-learning models improved receptivity by up to 36%. Receptivity to messages from the adaptive model increased over time.
Volume
27
Number
2
Start page
23
End page
28
Division(s)