AI-Driven Scheduling under Uncertainty for CT/MRI Operations: A Literature Review

Planning when to perform computed tomography and magnetic resonance imaging is complicated, because the initial schedule is seldom maintained with perfect accuracy during the course of a day. The duration of each scan is specific to the examination and the particular patient, and the total number of scans completed is affected by patients arriving late, appointments being missed, the need for emergency studies, and malfunctions of the equipment. Therefore, simply looking at how much the scanners are in use or whether all appointments are booked does not allow for a proper evaluation of scheduling efficiency. More important metrics are patient waiting times, the amount of overtime required, how firmly the schedule holds, and how many appointments must be changed.

For the purpose of this assessment, computed tomography and magnetic resonance imaging departments are being used to illustrate appointment systems with many sources of unpredictability: fluctuating examination lengths, patients who do not attend, emergency cases, equipment restrictions, and their reliance on hospital processes. This research review looks at studies of situations where the number of patients is not known in advance, examination durations vary, patients do not attend appointments, urgent scans are needed, or the usual procedures are disturbed. Importantly, it is not exclusively about filling available appointments. The review also considers how predictions and regression analysis are used to estimate the changeable parts of a department’s operation and whether scheduling plans are tested by simulation, digital twins or adaptive control before being used in practice.

The studies included indicate that predicting what will happen is only a single component of successful scheduling. Departments also need to decide on the appropriate amount of leeway to include, how many appointment times to reserve for emergencies, and a reasonable maximum for additional hours.