Ward-Specific Probabilistic Patterns in Temporal Dynamics of Nursing Demand in Japanese Large University Hospital:...

As global populations age, a looming nursing shortage is anticipated to become a critical issue. Charge nurses have the responsibility of optimally allocating nursing resources to ensure the quality of patient care during a shift. Therefore, an accurate estimate of nursing demand is crucial. However, the ability to forecast future nursing demand remains underdeveloped, mainly because the nature of nursing demand is highly individualized and does not follow a definitive pattern. In practice, the nursing demand is often perceived as unpredictable, leading to an ad hoc approach to staffing. The primary objective of our study is to demonstrate that longitudinal data analysis can reveal strong statistical regularities in the temporal dynamics of nursing demand. This approach not only provides new possibilities for efficient resource allocation but also paves the way for data-driven prediction of nursing demand. Our study uses Sankey diagrams to visualize the temporal dynamics of nursing demand within each ward for each fiscal year, representing these dynamics as an overlay of trajectories from multiple individual patients. Consequently, our study reveals ward-specific statistical regularities in the temporal dynamics of nursing demand. In one ward, approximately 25% of patients experienced an increase in nursing demand from 1 to between 6 and 9 points from the second to the third day of hospitalization, while in another, only 0.1% showed such an increase. These findings suggest that patients admitted to the wards tend to exhibit a certain probabilistic change in nursing demand. This study can predict probabilistically the temporal variation of nursing demand among patients in the coming years by analyzing data on the temporal changes in nursing demand over the past years. Our findings are expected to significantly influence the forecasting of nursing demand and the estimation of nursing resources, leading to data-driven and more efficient nursing management.