142 Chapter 5 The need for adaptability of interventions and implementation strategies has gained increasing attention in research, including the design of adaptive trials [51] and methods to track adaptations in implementation strategies [52]. The project team played an active role in communicating findings and adapting both the intervention and its implementation in response to determinants. These adaptations were carefully documented in field notes. The accessibility of DRAAI within the electronic health record was an important barrier, with nurses frequently asking how they could access it more easily. Integrating DRAAI within the electronic patient record minimizes disruptions of nurses' workflow, thereby enhancing the acceptability and feasibility of using a clinical decision support system [34]. Strengths and limitations A significant strength of our study was the collaboration among nursing scientists, data scientists, quality and care teams and health care professionals. From the development phase through the evaluation phase, close collaboration with data scientists facilitated rapid updates and data checks, allowing identified issues to be resolved within a single day. He, Baxter [53] highlights the critical role of such collaboration in ensuring the success of AI implementations in healthcare. The involvement of quality and care teams and nursing scientists fostered alignment between clinical practice and science. However, the involvement of the project team also presented certain limitations. While their embedded role enabled rapid adjustments to implementation strategies, it may have compromised objectivity and introduced the risk social desirability bias [54]. Additionally, the short interval between distributing the questionnaire and nurses’ exposure to DRAAI – ranging from 2 to 8 weeks – may have contributed to inconsistent responses. Some nurses reported not having used DRAAI themselves, though they may have encountered it during daily stand-ups or while observing colleagues. We conducted extensive evaluation sessions with the local implementation teams; however, these sessions were not recorded. Relying solely on field notes, without supplementing them with additional interviews or focus group discussions, limited our ability to gain a deeper understanding of the mechanisms of change that were addressed or overlooked in this pilot study. Recommendations A unique aspect of this study is the implementation of an AI-based prediction model in clinical practice, unlike many other models that fail to progress beyond the development phase. This success is largely due to our continuous re-examination of
RkJQdWJsaXNoZXIy MTk4NDMw