In healthcare, workload management is essential for providing stable and high-quality services. The method has been proposed to adjust nurse staffing as discrete workload management for the target of relatively long periods. However, changing circumstances surrounding operations are inevitable in healthcare, and mental workloads change in response to changing circumstances. To realize management to accommodate such changes, it is necessary to estimate mental workloads in real time and reflect them in management. However, no method has been established to estimate nurses’ mental workloads in real time according to changing circumstances. Therefore, in this study, aiming at the application for workload management, we construct a model to estimate an individual’s mental workload in real time. To acquire the data for a workload estimation model, we conducted a simulation experiment in a simulated intensive care unit. Using data recorded in the simulation experiment, we attempted to construct a model to estimate mental workloads using the indicators of nurses and ward conditions as features. The model could reflect the gradual changes in workloads, and the model would be expected to support workload management by providing feedback on the nurses’ mental workload.