Digital Mental Health & Explainable AI
Date: Monday 10 Nov 2025
Room: S/2.22 - Sandpit/Katherine Johnson Suite
| Time | Presentation | Abstract | Authors | Contribution type |
|---|---|---|---|---|
| 11:00-11:20 | Culturally Adapted Design of a Digital Mental Health Intervention to the Chinese context: A design case study | Culture plays a crucial role in the design of mental health interventions since it influences how people seek assistance, participate in healthy behaviours, and how services are provided. It is believed that by including cultural factors into the intervention, the relevance, acceptability, effectiveness, and sustainability would be improved. In this paper, we present the design process of a case study of adapting an evidence-proven Australian digital mental health intervention to the Chinese context, following culturally sensitive design frameworks including the ADAPT Model and the Ecological Validity Model. Through cultural adaptation, the localised intervention, namely 云舒 (CloudEase), received a higher overall satisfaction score in the System Usability Scale, compared to the literal translated version of the original intervention. Documenting each step of this process demonstrates a practical roadmap and guidelines for customising similar digital mental health interventions in Chinese or other cultural settings. | Sijin Sun, Jiyuan Cao, Zheyuan Zhang and Rafael A. Calvo | Research |
| 11:20-11:40 | Co-Designing Augmented Paper for Health Education with Older Adults | Health education plays a crucial role in promoting healthy ageing. This study investigates the co-design of two augmented paper prototypes to enhance health education provision for older adults. Two seed concepts were developed based on user requirements from prior research: (1) self-printed postcards for physical exercises linked to audio or video demonstrations on a smartphone, (2) a folding leaflet on eating well linked to explanatory text, audio and video clips. Links were implemented through a custom app (1) or a QR LinktreeTM (2). Ten participants aged 65 or older were invited to re-design each concept, using the Focusgroup+ method. Two new concepts revealed a shift of printed text content from paper to audio, and an innovation of self-constructed printed paper formats following medical consultations or online health information searches. These demonstrate the potential for blending physical and digital resources to enhance the accessibility of health information for older people. | Larissa Taveira Ferraz, David Frohlich, Charo Hodgkins and Paula Castro | Research |
| 11:40-12:00 | Are All Design Frictions Equal? Exploring Types of Frictions and their Perceived Value for Digital Wellbeing | Much research on digital wellbeing has focused on interventions for digital self-control while less work explored design frictions (which deliberately aim to inconvenience users) despite their potential to more sensitively account for users’ contexts. To address this gap, we identified five types of design frictions: cognitive, emotional, motivational, social, and physical for which we provided working definitions, and illustrated with design exemplars from the state-of-the-art. We also report workshops with 19 participants to explore these types of frictions and users’ perception of their value for digital wellbeing. We conclude with four design implications for digital wellbeing including supporting design frictions at app level, supporting adaptive design frictions, sensitive design of emotional frictions, and balancing user’s need for privacy in social frictions. | Sultan Almoallim and Corina Sas | Research |
| 12:00-12:20 | Expert Nutritionists’ Perception of XAI Visualisations for Overweight Prediction | The integration of Artificial Intelligence (AI) in overweight prediction has shown significant promise. However, the adoption of these systems relies healthcare professionals’ understanding of AI algorithms and outputs. This study explores how expert nutritionists understand and engage with AI Random Forest Classifier algorithm and understand its prediction while using visualisations from two Explainable AI (XAI) tools: SHAP LIME. We report workshops with ten practitioners to investigate also their perception of the dataset used for overweight prediction, and their experiences with SHAP and LIME visualisations. Our findings highlight insights regarding dataset provenance and relevance, benefits and limitations of these visualisations and key features supporting nutritionists’ understanding of overweight prediction. We conclude with a reflection on our key findings and their importance for AI-HCI research in health. | Ahmad Alaqsam and Corina Sas | Research |