Utilization of EMA and the Transtheoretical Model of Behavior Change to Prevent Relapse during Long-Term Socially Assistive Robot Based Interventions
—Longitudinal research is fundamentally important for understanding sustainable behavior change related to human health. This is particularly true for chronic illnesses, such as dementia and mental illness, where treatment is often measured in months and years over the long-term. For research on human-robot interaction (HRI) geared towards addressing such health-related issues, this presents a challenge in that many HRI studies (particularly in lab settings) are performed in a matter of minutes or at most hours. Given that reality, this suggests a need for innovative approaches to longitudinal research in HRI for such health-focused scenarios. Here, we argue that such methodological approaches to longitudinal HRI research should be grounded in existing models of behavior change (e.g. transtheoretical model) and leveraged via novel technology-enabled real-time sampling strategies for user activities outside of lab settings (e.g. ecological momentary assessment, EMA), all geared toward supporting data-driven methods for connecting robot behavior and human behavior over the long-term in in-the-wild settings.