Challenges in Change Management for AI-Driven Prediction Tools in Public Hospital Clinical Wards: The Case of mROC Implementation

Abstract

The integration of artificial intelligence (AI) in healthcare offers significant opportunities to enhance clinical decision-making and patient outcomes. However, AI adoption in public hospital settings presents various challenges, particularly concerning clinician resistance, concerns over predictive accuracy, and the perceived threat to traditional clinical judgment. This study examines the implementation of the Modelling Risks and Outcome Calculations (mROC) tool, an AI-driven predictive system developed to reduce hospital-acquired complications (HACs) at North Metropolitan Health Service (NMHS). The mROC pilot demonstrated notable success, particularly in reducing urinary tract infections (UTIs) within a neurosurgery ward, achieving a 48% reduction in UTI rates per 1,000 bed days and an estimated cost savings of $AUD406,758 over three months. Despite these promising results, significant barriers hindered broader implementation. Resistance stemmed from the disruption of established clinical workflows, scepticism regarding AI-driven predictions, and concerns about increased scrutiny over clinical decision-making. Furthermore, disparities between mROC risk assessments and traditional clinical assessments generated uncertainty about the tool’s reliability. Limited clinician engagement in the tool’s development also contributed to reluctance in its adoption, emphasising the importance of co-design in AI integration. This paper identifies key lessons from the mROC implementation, highlighting the necessity of early clinician involvement, transparent communication of AI effectiveness, and strategies for aligning AI tools with clinical workflows. Recommendations include structured change management approaches, iterative pilot trials, and improved real-time adaptability of AI models to evolving patient conditions. By addressing these challenges, AI-driven tools like mROC can foster sustainable adoption, optimising patient care while supporting clinical decision-making in public hospital settings.

Ähnliche Arbeiten