ZzzMate: Designing an Empathetic Chatbot for Addressing Self-Conscious Emotions for Sleep Adherence

Empathetic design is increasingly recognized as vital for behavioral interventions. However, self-conscious emotions like guilt, shame, embarrassment, and pride are often overlooked. We introduce ZzzMate, a mobile chatbot that integrates a model capable of detecting self-conscious emotions based on user input. These emotions, along with sleep goals and persuasive, empathetic strategies, are fed into a large language model (LLM) to generate tailored messages promoting better bedtime adherence. A three-week field study with 27 participants compared ZzzMate against a Baseline LLM and a simple notification system. Results showed that ZzzMate significantly improved user engagement, motivation, self-efficacy, and bedtime adherence, while also demonstrating superior emotional intelligence. In-depth interviews provided insights into users’ experiences with the system. Based on these findings, we propose design implications to guide the development of future behavioral change technologies, advocating for empathetic systems that resonate with users’ nuanced emotional states, ultimately contributing to a more humanized technological future.