Workshop for Junior Researchers: The Transformative Potential of Generative AI: Three Examples from Digital Health Research

The Transformative Potential of Generative AI: Three Examples from Digital Health Research

Abstract:

The transformative potential of Generative AI (GenAI), particularly Large Language Models (LLMs) such as ChatGPT, is significantly reshaping digital health research and practice. In this presentation, we will showcase three concrete examples from our research at the Ludwig-Boltzmann Institute for Digital Health and Prevention in Salzburg, Austria. These examples showcase our core research pillars in digital health: Assessment, Guidance, and Motivation.

Focusing first on assessment, Pavithren Pakianathan will discuss how integrating LLMs into multi-modal data interpretation can support healthcare professionals during cardiac rehabilitation planning sessions. He will share insights and preliminary findings from ongoing research, illustrating how AI can improve sensemaking of patient-generated health data, potentially improving shared decision-making.

Under the umbrella of guidance, Faith Young will present an innovative mobile application developed to encourage physical activity, turning short breaks and waiting periods into exercise opportunities. The app employs an AI model that generates contextually relevant exercise suggestions based on users’ photographs of their surroundings. Faith will outline the application’s design and the methodology of an upcoming study assessing its effectiveness compared to a control condition featuring generic, predefined exercise recommendations.

Finally, David Haag and Dominik Hofer will examine the potential of LLMs in revolutionizing Just-in-Time Adaptive Interventions (JITAIs) aimed at promoting physical activity. They will discuss results from a formative study comparing the acceptability and effectiveness of LLM-generated JITAIs with those designed by healthcare professionals and laypersons, addressing both opportunities and technical challenges in leveraging LLMs for personalized interventions.