Quantitative Insights From Nursing Workload Measurement: An Artificial Intelligence-Driven Approach.

INTRODUCTION: Military nurses face a unique set of challenges compared to their civilian counterparts, including frequent deployments and high workloads because of shortages of trained military nurses. However, quantifying nurse workload is difficult. Advances in informatics and electronic health record (EHR) audit log data show promise in objectively and unobtrusively measuring clinician work. Through an exploration of audit log data, we demonstrate that this rich source of temporal data that can represent nursing activities, workload, and workflow patterns. MATERIALS AND METHODS: Five years of audit log data were extracted from an academic medical center for all nurse-user roles, capturing timestamps, and EHR actions. We applied temporal mining techniques to reveal patterns in nursing workflows and to identify high-intensity EHR interaction periods. For future use, we explored the adequacy of predictive modeling techniques to assess workload variations and forecast nurse burnout risks. RESULTS: We found that 8,149 unique nurse users logged 1,461 distinct types of EHR activities (tasks). Many of the tasks consisted of viewing or searching for EHR data. By organizing audit log data in a way that facilitates time-sequenced analysis, we created Predictive Optimization of Workload and Efficiency for RNs (POWER), a reproducible framework that can be used for automated AI-driven workload monitoring. CONCLUSIONS: Although this work focus on nursing logs, this methodology is much more general and demonstrates how audit log data can be leveraged for scalable and unobtrusive workload analysis, offering a more objective and automated approach. Next steps will focus on validation in military hospitals, co-design of workload visualization tools, and exploring AI-driven decision-support tools to recommend optimized staffing models and workflow adjustments, paving the way for AI-driven nursing workforce management.