1. Human centered design of AI-powered Digital Therapeutics for stress prevention: Perspectives from multi-stakeholders' workshops about the SHIVA solution
- Author
-
Marco Bolpagni, Susanna Pardini, and Silvia Gabrielli
- Subjects
Digital Therapeutics (DTx) ,Human-centered design ,Affective computing ,Stress prevention ,Stakeholders engagement ,Virtual reality ,Information technology ,T58.5-58.64 ,Psychology ,BF1-990 - Abstract
Background: AI-powered Digital Therapeutics (DTx) hold potential for enhancing stress prevention by promoting the scalability of P5 Medicine, which may offer users coping skills and improved self-management of mental wellbeing. However, adoption rates remain low, often due to insufficient user and stakeholder involvement during the design phases. Objective: This study explores the human-centered design potentials of SHIVA, a DTx integrating virtual reality and AI with the SelfHelp+ intervention, aiming to understand stakeholder views and expectations that could influence its adoption. Methods: Using the SHIVA example, we detail design opportunities involving AI techniques for stress prevention across modeling, personalization, monitoring, and simulation dimensions. Workshops with 12 stakeholders—including target users, digital health designers, and mental health experts—addressed four key adoption aspects through peer interviews: AI data processing, wearable device roles, deployment scenarios, and the model's transparency, explainability, and accuracy. Results: Stakeholders perceived AI-based data processing as beneficial for personalized treatment in a secure, privacy-preserving environment. While wearables were deemed essential, concerns about compulsory use and VR headset costs were noted. Initial human facilitation was favored to enhance engagement and prevent dropouts. Transparency, explainability, and accuracy were highlighted as crucial for the stress detection model. Conclusion: Stakeholders recognized AI-driven opportunities as crucial for SHIVA's adoption, facilitating personalized solutions tailored to user needs. Nonetheless, challenges persist in developing a transparent, explainable, and accurate stress detection model to ensure user engagement, adherence, and trust.
- Published
- 2024
- Full Text
- View/download PDF