Research Overview

In the nursing field, the “duty roster” is often referred to as the “nurses’ bestseller” because it holds the key to a nurse’s entire month. As soon as the roster is released, nurses rush to check it, confirming whether their requested days off have been granted, who their night shift partner will be, and whether they have been assigned leadership duties—reacting with a mix of joy and disappointment depending on the results. These emotional reactions are not merely fleeting feelings; they are deeply personal and directly impact each nurse’s “quality of life.” On the other hand, wards have a social responsibility to provide appropriate nursing care 24 hours a day, and ensuring “quality of care” is essential when creating duty rosters. Balancing these two critical elements—quality of life and quality of care—while planning within complex constraints places a significant physical and mental burden on ward nursing managers. This study explores methods for creating high-quality work schedules while alleviating the burden faced by frontline managers.

What is Nurse Scheduling?

The creation of nursing schedules is a subject of such active research that it has become a distinct field known as the “nurse scheduling problem.” However, in many cases, this term refers broadly to “schedules that are difficult to create,” with the focus primarily on optimization methods and the development of generation algorithms. There is also a significant amount of research on schedules in other fields, and a key characteristic of these findings is that they are not limited to the nursing field, which makes a broad contribution to society. Furthermore, there are numerous software tools available that support the creation of work schedules using automated generation, a key research outcome. However, these tools do not fully meet managers’ expectations; even when implemented, the generation functions are often unused, and the software is typically used only as a work schedule editor for inputting and editing data. This trend is commonly observed in nursing settings in Japan. Although this problem has been identified for over 20 years, a fundamental solution has yet to be found.

Reflections on the Ideal Form of Software

Creating nursing schedules cannot be solved simply by using software to “automatically generate” them with a single click. We believe it is difficult to successfully incorporate and account for the complex “conditions” that exist on the front lines—such as unwritten rules that are hard to quantify, the ever-changing dynamics of interpersonal relationships, and personal circumstances—into the generation process. In this study, we do not view the role of software as simply “performing tasks in place of humans.” Rather, we are constantly considering how it can serve as a “supportive partner” that assists the creator’s “judgment and thinking.” One approach to this partnership involves “quantitatively evaluating” the duty rosters created by managers. For example, we assess the sequence of shifts, the distribution of days off, and the feasibility of the schedule.

Exploring quantitative evaluation methods

Some aspects of quantitative evaluation are difficult to quantify. Examples include “team evaluations” and “tacit knowledge.” These rely heavily on the manager’s subjective judgment. To quantify this subjectivity, we are attempting the following approaches. For team evaluations, we start from the premise that while evaluating large teams is difficult, pair evaluations are feasible when referencing social networks. We attempt to quantify these by following the process of “pair evaluation” → “graph (network) creation” → “graph evaluation.” Additionally, for tacit knowledge, we are attempting to identify the “hidden conditions” embedded in the work schedules created by managers by leveraging data science. We believe that exploring quantitative evaluation methods will not only help us communicate the merits and demerits of work schedules to managers but will also lead to training for new managers on how to create work schedules and, in the future, support for AI-driven work schedule creation.