Engaging Hospital Staff to Identify Levers for Clinical Decision Support Adoption: System Dynamics Group Model Building (Preprint)

BACKGROUND Clinical Decision Support tools that provide patient-specific and evidence-based information to providers and care-managers regarding patient risk for adverse outcomes have been a part of health care for decades. However, modern clinical decision support, which consist of automated predictions based on complex machine learning models and hundreds of complex input variables, faces obstacles to adoption related to providers’ perceptions of lack of transparency and utility. Often the expertise of data scientists and clinical end users are nt well-integrated, creating implementation gaps from clinical decision support development to adoption and ongoing implementation. OBJECTIVE We developed this protocol to enlist group model building from system dynamics. GMB activities bring multidisciplinary stakeholders together to identify facilitators and barriers to implementation of one class of clinical decision support in general medical-surgical wards, early warning scores. By modeling the adoption and implementation of the early warning score clinical decision support in general medical-surgical wards, we have sought to evaluate GMB activities as a generalizable strategy for improving clinical decision support adoption and implementation in other contexts. METHODS We have worked with health system interest holders to co-develop a model of facilitators and barriers to early warning score adoption in workshops that apply System Dynamics Modeling techniques of eliciting key variables, describing their behavior over time, and producing a diagram that illustrates the stock-and-flow and feedback dynamics of early warning score adoption and implementation over time. RESULTS The end result will consist of an initial demonstration model that formally depicts dynamic hypotheses about how system behavior affects clinical decision support early warning score uptake. CONCLUSIONS Completion of these aims will advance knowledge about the practice of improving machine learning-enabled clinical decision support adoption and implementation using group model building.