Improved arithmetic optimization algorithm for patient admission scheduling problem

The patient admission scheduling problem (PASP) has been studied for many years as one of the most important scheduling problems in the health sector. The primary goal of PASP is to assign patients to appropriate hospital beds while considering some hard and soft constraints. PASP is an NP-hard problem, which implies that the optimization approach is one of the best approaches that can be used to solve it. The arithmetic optimization algorithm (AOA) is a new optimization algorithm that can effectively solve both continuous and discrete optimization problems. However, it may suffer from poor exploration and premature convergence to sub-optimal solutions due to some problems with its exploration operators. In this paper, we propose a new method for the PASP using an improved AOA algorithm called the island-based AOA (iAOA). The new algorithm is based on a structured population model called the island model. This model distributes the population of candidate solutions among islands that periodically exchange some of the candidate solutions among each other based on a migration protocol. In iAOA, some modifications were applied to AOA’s parameters to make it capable of handling discrete space. We evaluated iAOA using a public benchmark for the PASP and compared our results with those of other baseline algorithms (DFPA, SA, HS, GD-NLGD, HTS, TS, I-BBO, and MBBO). The simulation results revealed that iAOA achieved the minimum average execution time on 5 out of 6 instances of the public benchmark. It also showed that iAOA achieved the second minimum average violation of the objective function over 10 independent runs. Finally, the statistical examination of the experimental results using Friedman and Wilcoxon tests confirms the significance of the results of iAOA compared with the other baseline algorithms.