Trusting a Machine? Opportunities and Challenges to Using Artificial Intelligence in Qualitative Patient Research

Issue: Qualitative patient research is at the center of patient-focused drug development, contributing to the selection and development of endpoints that evaluate new therapies on outcomes that matter to patients. However, robust qualitative research takes time and requires the use of highly experienced staff. AI has the potential to contribute to efficient and scaled delivery of qualitative research to ensure patient voices can be reflected in a wide range of decisions across the product life cycle. Overview: This session will delve into the practical and ethical issues raised when introducing the use of AI into qualitative data analysis from the perspectives of study sponsor and researcher, drawing on practical experiences. The debate will be guided by the practical experiences of the panel members, including a pilot study that compared human and AI tasks in the analysis of qualitative data for regulatory submissions. Christina Silver will moderate the session and provide a brief history of the use of AI in qualitative analysis leading to the present-day. Karen Bailey and Paul Cordero will discuss a case study of AI use-case with data from a hybrid concept elicitation and cognitive interview study involving non-Hodgkin’s lymphoma patients in order to understand how the use of AI can support the quality and efficiency of outputs. The speakers will reflect on the value of AI with human interaction in the development and use of the codebook, qualitative analysis plan, coding, content analysis, table outputs, and report writing. They will present the key scientific and potential regulatory challenges and preliminary guidance for use-cases for AI analysis of patient qualitative data. All speakers will provide practical insights from real applications but will also debate unsolved issues that require consensus among the sponsors, regulators, payers, scientific community and wider society.