Machine Learning Augmented Handoff Report
In efforts to adhere to the National Patient Safety Goals (NSPG) for handoff standardization, EHR-based handoff templates are designed as a “one-size fits-all” solution. However, these rigid tools fail to: (a) address the handoff needs and workflows unique to the different healthcare disciplines in the various settings, (b) tailor to patient cases and related risks; (c) foster the core functions of handoffs such as information processing, distributed cognition, common ground, and anticipatory management.
We propose to address these failures within the context of postoperative handoffs (operating room to intensive care unit). In this study, our goal is three-fold: First, use machine learning (ML) techniques to predict postoperative patient risks by harnessing preoperative and intraoperative real-time data from EHR. Second, adopt a user-centered design approach to develop a flexibly-standardized, user-friendly handoff report. Third, incorporate features in the handoff report that supports the core functions of handoffs.