Preliminary survey findings on PharmaBot – a retrieval-augmented chatbot for insulin safety information and pharmacovigilance support

Adverse drug reactions (ADRs) pose a significant patient-safety concern, contributing to substantial morbidity, mortality, and healthcare utilization. Source-grounded chatbots have the potential to enhance pharmacovigilance by improving access to regulated medicine information and reducing barriers to ADR reporting. We created a prototype of a web-based chatbot (PharmaBot) designed using retrieval-augmented generation methods to provide responses grounded in summaries of product characteristics and package leaflets. Subsequently, we conducted a cross-sectional survey among 37 participants who were treated with insulin to assess their digital literacy, perceived usefulness, trust, and intended use of the chatbot for drug safety information. The results indicated that most respondents (70.3%) demonstrated high to very high smartphone app skills. Moreover, 73.0% rated the chatbot as useful or very useful, and 78.4% expressed moderate to very high levels of trust in it. About 56.8% of participants reported an intention to use the chatbot regularly (frequently or very frequently). The most popular intended use cases were therapy reminders (81.1%) and information about insulin and therapy (70.3%), while 51.4% chose guidance on ADR reporting. These preliminary findings suggest that users have an acceptable perception of a source-grounded chatbot as a supportive tool for pharmacovigilance-related workflows.