Creating an optimal OR schedule regarding downstream resources

Creating an optimal OR schedule regarding downstream resources

Carlier, M.

Van Essen, T. (mentor)

Electrical Engineering, Mathematics and Computer Science

Applied Mathematics

2017-03-06

A high percentage of hospital admissions is due to surgical interventions. The operating theatre, which holds the operating rooms (ORs), is therefore one of the key resources in hospitals. Managing the operating theatre and finding an optimal OR schedule is complex due to the many factors that influence it. Scheduling a surgery in an OR influences downstream facilities like the post anaesthesia care unit, intensive care unit and general patient wards. This research was conducted at Leiden University Medical Centre (LUMC), an academic teaching hospital in Leiden, the Netherlands. During the week, the LUMC experiences a large variation in bed occupancy at the patient wards due to the way surgeries are scheduled. The large variation in bed occupancy causes surgeries to be cancelled, because there are no beds available at the ward. Because the OR theatre is such an expensive resource, we want to find a schedule that utilises the OR optimally during opening times. In this research, we develop a clustering method to cluster surgical procedures into surgery groups based on surgery duration and the length of stay. Then, we extend a model that analytically determines the patient distributions over the wards and intensive care for a given OR schedule. We define a mixed integer programming model with the objective to maximise the OR utilisation and minimise the variation in bed occupancy at the wards and intensive care. The model produces an OR schedule with the defined surgery groups assigned to days in the OR. We use two different methods to solve the model: a global approach and a local search heuristic, i.e., simulated annealing. The model has one nonlinear constraint and a complex objective function. Therefore, we linearise the constraint and the objective function, which results in a mixed integer linear program that is proven to be 𝑁𝑃-hard. Both approaches are tested on a dataset provided by the LUMC. Furthermore, several scenarios are evaluated. We conclude that the mixed integer linear programming method performs better and faster than the simulated annealing procedure. To obtain an even better solution it is possible to use a combination of both. By using this method, the OR utilisation of the LUMC can improve by 11% and the variation in bed occupancy can be decreased by 80%.

master surgery schedule
Operating room scheduling
bed occupancy
mixed integer linear programming
simulated annealing
length of stay
optimization
hospital

http://resolver.tudelft.nl/uuid:69719e2d-5649-47da-a39e-e9107487eab1

Student theses

master thesis

(c) 2017 M. Carlier