On June 29, 2026, the School of Media and Communication (SMC) at Shenzhen University hosted an academic lecture on patient-centered AI design for mental health support. The lecture was delivered by Professor Sebastian Scherr of the University of Augsburg, Germany, and moderated by Professor Bolin Cao, Director of the Department of Internet and New Media at SMC. Faculty members and students attended the event.
Professor Scherr opened the lecture by highlighting the persistent shortage of mental health resources and the growing potential of conversational AI to provide accessible, personalized, and empathetic support. AI chatbots, he noted, can offer around-the-clock assistance at relatively low cost, but their effectiveness depends heavily on how users experience the interaction. Poorly designed systems may lead to repetitive conversations, limited functionality, and declining user trust.
Against this backdrop, Professor Scherr introduced his research on how users evaluate different combinations of communication features in AI chatbots for mental health support. Rather than examining individual AI features in isolation, the research focuses on how multiple communication attributes work together and which features users prioritize when trade-offs are required.
He then presented two empirical studies. The first applied a patient-centered communication framework and used best-worst scaling with 414 participants in the United States to identify the communication attributes users valued most. Building on these findings, the second study employed a discrete choice experiment involving 1,011 participants, 268 of whom were assigned to an AI chatbot scenario, to estimate the relative weight users assigned on different combinations of communication features. The findings showed that symptom assessment and attentive listening were the two attributes participants valued most when seeking mental health advice. These were followed by the ability to explain medication-related information and provide community-based assistance. Notably, even when users preferred highly personalized communication, symptom assessment and attentive listening remained central to their choices.
During the Q&A session, faculty members and students engaged Professor Scherr in a lively discussion on AI ethics, user trust, and the challenges of translating AI-based mental health support into clinical practice. Professor Scherr emphasized that effective mental health AI should be designed around users’ needs rather than technological sophistication alone. The lecture offered participants new perspectives on the intersection of digital health communication, patient-centered design, and human-AI interaction.
