From Development to Clinical Impact: The Role of the User Experience of Artificial Intelligence Models for Clinical Decision Support

Abstract

This research investigates how the user experience (UX) of AI models influences healthcare providers' decision-making processes and identifies actionable insights for better clinical implementation. The findings indicate that the current UX of three specific AI models has minimal impact on decision-making, although healthcare providers find the models useful and are willing to use them. This suggests that improved UX could enable these AI models to significantly influence clinical decisions. The study highlights the importance of proper introduction, support, context, data load balance, and training in optimizing the UX of AI Clinical Decision Support Systems (AI-CDSS). Effective use of AI models is highly context-dependent, and further research is needed to identify optimal contexts. While there is a general preference for interfaces with balanced information, the inconclusive results on data load underscore the need for additional research. The research emphasizes that although the current UX of these AI models for the PICU and NICU of the WKZ is not yet fully optimized, addressing the identified aspects could enhance their contribution to clinical decision-making. By focusing on UX, the study provides valuable insights into bridging the gap between AI model development and their actual clinical impact, underscoring the critical role of UX in healthcare AI models.