Search

Technology based interventions can be used to address racial health disparities among patients with heart failure (HF) (Hughes & Granger, 2014). To avoid unintended consequences of tech-solutions developers need to understand how users interact with them and account for disparities in existing systems (Valdez & Brennan, 2017) To test HFChat, a spoken interactive dialogue agent, and understand how Black/African American (AA) and Hispanic/Latino (H/L) patients with HF interact with tech-based health solutions. HFChat is an interactive dialogue agent to assist AA and H/L participant with HF by answering self-care questions outside of the clinic. We developed the first iteration of HFChat based on current literature and findings from our analysis of 20 education sessions between health educators and patients with HF (18 AA + 2 H/L). To pilot test HFChat we conducted semi-structured interviews assessing the participants baseline knowledge about their heart condition, their self-care needs and routines. At the end of the interview participants (12 AA + 2H/L) were introduced to HFChat and allowed to interact with it. • ∼35% of the 14 participants couldn't name their heart condition despite an official HF diagnosis for 1+ years. When interacting with HFChat, these 5participants asked 2 questions on average (Box 1) while the remaining participants asked 5 on average (Box 2.) • Participants who could name their heart condition also asked about topics not covered by the current version of HFChat e.g., comorbidities, impact of different food groups on heart health etc. • Patient questions differed across users, with some asking succinct, direct questions, while others required prompts from HFChat. (Box 1 & 2) Please note that the Patient questions in Box 1 & 2 are as literally understood by HFChat, which explains some awkward phrasing and HFChat robustness to imperfect input. Patient usage of tech-based solutions differs by a number of different variables. Multiple identities (age, gender, tech-literacy, etc.) an individual subscribes to impacts how users interact with technology (Valdez & Brennan, 2017). Our preliminary work here demonstrates that solutions like HFChat need to account for a wide range of patient experiences and knowledge to be successful. [ABSTRACT FROM AUTHOR]