Publication
The influence of dialogue flow on stress levels when booking healthcare appointments with AI
Milos Kravcik; Elisabeth Reiswich; Iwan Lappo-Danilewski; David Buschhüter; Patrick Jähnichen
Artificial Intelligence in Healthcare Collection of Short Abstracts, Zenodo, 9/2025.
Abstract
Advancements in artificial intelligence (AI) have opened new possi-bilities for user adaptation, providing greater accessibility to healthcare and men-tal health support. Stress is an inevitable phenomenon in modern society and has become a critical factor for well-being in both the professional and personal spheres. This paper examines the application of human-centred AI in an appoint-ment-booking scenario, with a focus on stress detection in patients. The study utilises pre-trained machine learning classifiers for user adaptation and examines the effect of various prosody rates on stress, considering measurements to iden-tify physiological biomarkers associated with psychological stress. Test scenar-ios involving appointment booking, prescription and referral requests via the dig-ital phone assistant were explored. The speech rates of the AI assistant were ran-domised and varied from slow to normal to fast. Our analysis reveals promising results in distinguishing between static and dynamic systems, using a sample size of n = 12 in a user study. Stress responses measured with Empatica EmbracePlus suggest that dynamic systems are preferred over static ones. This finding could be replicated using self-reports from participants in the study. This work contrib-utes to the growing body of research on digital health tools for healthcare assis-tance, highlighting the need for interdisciplinary collaboration to advance the field responsibly. With stress-related disorders and AI usage rising globally, un-derstanding the interaction between stress and automated dialogue flow could provide helpful strategies to improve the user experience, which could then be scaled up to other health and work environments.