Co-design of a Conversational Artificial Intelligence Program to Support Self-management in Heart Failure Patients

Abstract

Heart failure (HF) is a leading cause of hospitalisations and mortality worldwide. Effective self-management has been shown to improve health outcomes, however, remains challenging for patients with HF. Digital technologies offer a novel way to support these patients, but engagement remains suboptimal. Previous studies have highlighted the potential of 'chat-like' features in increasing patient engagement.   This study aims to co-design a prototype for CardioChat, a conversational artificial intelligence (AI) intervention to support HF self-management. This will be conducted following the Generative Co-design Framework for Health Innovation, comprising 3 phases: pre-design, co-design, and post-design. The 'pre-design' aims to identify the self-management needs of patients with HF and explore the potential of this technology in supporting them. It will be done by conducting a systematic review and meta-synthesis, and through a focus group with 4-6 multidisciplinary clinicians experienced in HF management, and 1-2 engineers knowledgeable in conversational AI. The 'co-design' aims to shape the content for CardioChat. It will be done through an interactive workshop with 10 patients with lived experience of HF using a persona-scenario exercise to gather insights for creating content, including scripting interactions that could support them with daily self-management tasks. Focus groups will be analysed using thematic analysis, while workshop outcomes will undergo directed qualitative content analysis. A prototype will be built based on these findings, which will be refined in the 'post-design' through a second focus group with clinicians and engineers. The final prototype will then undergo user acceptance testing with HF patients.