Designing an AI-Driven Chatbot to Address Radiographers’ Needs: A Mixed-Methods Approach

Aim

Generative AI is revolutionizing industries by streamlining workflows and facilitating informed decision-making. However, existing AI-driven chatbots often target general audiences, offering broad functionalities that do not address the specific needs of niche professions. In healthcare, AI has demonstrated promise in automating tasks and improving knowledge retrieval, yet its application in radiography remains underexplored. Radiographers face unique challenges, including accessing updated imaging protocols, navigating diverse workflows, and making rapid decisions under time constraints. This study aims to bridge these gaps by assessing radiographers’ challenges and exploring their expectations of AI-driven chatbots to inform the development of a tailored solution that enhances efficiency and decision-making.

Methods

An explanatory sequential design was employed, beginning with a survey of radiographers (n=39) to identify key challenges and expectations. Quantitative data were analysed using descriptive and inferential statistics. Subsequently, focus group discussions (FGDs) were conducted with purposively sampled participants (n=9), and transcribed through verbatim, and thematic analysis was conducted to gain deeper insights into radiographers’ experiences, perceptions, and expectations.

Results

Survey findings revealed that 87.2% of radiographers struggled with accessing updated imaging protocols, while 59% faced difficulties navigating workflows across different environments. Participants strongly prioritized quick access to imaging protocols (87.2%) and tailored workflow guidance (84.6%). Concerns about chatbot accuracy (87.2%) and ease of use (43.6%) were also prominent.

Thematic analysis revealed that fragmented resources and rotational roles were key factors underlying these challenges. Participants expressed a need for AI-driven chatbots to provide quick and reliable decision support, prioritizing features like rapid access to imaging protocols and workflow guidance. Chatbots were seen as a promising solution to standardize practices, reduce reliance on colleagues, and enhance efficiency. Critical factors influencing adoption included trust in the chatbot’s accuracy, evidence-based recommendations, concise responses, and mobile accessibility, enabling seamless integration across diverse clinical settings.

Conclusion

This study identifies radiographers’ operational challenges and provides actionable insights into their expectations for an AI-driven chatbot. A tailored solution that centralizes updated protocols, enhances decision-making, and supports workflow navigation can improve efficiency and reduce stress among radiographers. Future work will focus on integrating these findings into chatbot design and evaluating its impact on clinical workflows and patient care outcomes.