Human-computer interactions and compassionate healthcare: A Wizard of Oz study using a self-administered AI-assisted cognitive assessment

IntroductionArtificial intelligence (AI) is increasingly transforming healthcare, however, the interaction between the user, AI, and the users environment is poorly understood. To elucidate this interplay and support the delivery of compassionate remote care in Occupational Therapy (OT) practice, we created a self-administered remote AI-powered cognitive assessment, facilitated using the Wizard of Oz method, and measured the experiences of patients, controls, caregivers, and healthcare providers. The Wizard of Oz method uses human-computer interaction (HCI) and user experience (UX) research to simulate the functionality of a system before it is fully developed by creating the illusion of a functioning system by having a human behind the scenes, controlling the systems responses. Research Questions: 1. How are AI-assisted cognitive assessments experienced by patients, caregivers, healthcare providers, and controls? 2. How can we maintain compassionate care while incorporating technology into healthcare? 3. What are the concerns of healthcare providers using chatbots to administer cognitive assessments? Methods6 participants with progressive cognitive decline, 6 healthy controls, 6 caregivers, and 6 healthcare providers were invited to complete a virtual AI-powered cognitive assessment followed by a survey about their experience and other demographical information. Survey questions were separated into 4 scales: Trust, Compassion, Usability, and Care Experience. ResultsNo statistically significant difference in mean survey scores between participant categories was observed. Factors such as sex, device type, chatbot familiarity, and education had no statistically significant effects. Participants scored statistically significantly lower on the scale Trust (8.09) than on Compassion (8.72). Additionally, those who used the chatbot during the assessment scored statistically significantly lower on the Usability scale compared to those who did not (7.33 vs. 9.20). ConclusionThe findings help to evaluate user experience with virtual AI-based cognitive assessments and provide insights that can inform important design characteristics to improve user experience and compassionate care delivery. Author SummaryAs more technological advancements are being achieved in our modern lives today, we continue to see an increasing amount of these technologies in our healthcare system as well. AI is a popular category of these technological advancements and appropriately implementing this powerful tool into day-to-day medicine may prove to benefit our healthcare system. However, we first need to understand our current views on how AI can affect our delivery of care. In this paper, we explored user experience specifically in the context of a cognitive screening tool using wizard of Oz methodology. In brief, our research explores how various stakeholders (patients, control, caregivers, healthcare providers) experience and feel about using a virtual AI-assisted platform for conducting cognitive assessments. We believe that further exploration of AI in medicine and how it can be improved provides an overview of our attitudes towards implementing artificial intelligence into our healthcare system and will also inspire further research for artificial intelligence in medicine.