Bi-Weekly: A Multiobjective Approach to Hospital Patient to Room Assignment
Wednesday, October 17, 2018, 10:30am
Location: RWTH Aachen University, Department of Computer Science - Ahornstr. 55, building E3, room 9u10
Speaker: Tabea Krabs
Patient to room assignment is an important task in hospitals as it has a major impact on the degree of capacity utilisation and care quality. The goal of patient to room assignment is to find an assignment that allocates every patient to exactly one room for every day of their stay respecting the room capacities and fundamental constraints such as necessary medical equipment or the specialism of the corresponding department. Naturally, this assignment should avoid transfers of patients between different rooms during their stay as much as possible. Furthermore, assignments of female and male patients to the same room at the same time are not desirable. We propose a new interpretation of patient to room assignment as a multiobjective optimization problem, where we consider the suitability of an assignment, the total number of transfers and the total number of gender mixed room assignments as separate objective functions rather than combining them with predefined weights to a single objective function. This allows more flexibility in decision making than formulations that are based on predefined proportions of the different cost factors and goals. In this talk, we present results on the complexitiy of this problem and some structural insights of the Pareto front. Furthermore, we compare different approaches to solve this multiobjective optimization problem such as scalarization approaches or heuristics for computing the Pareto front.