Multi-Objective Physician Scheduling using Non-Dominated Sorting Native Binary Particle Swarm Optimization with Deallocation Method

Abstract

Scheduling Optimization is a critical operation in Hospital Management. A good medical personnel schedule can overcome a lot of troubles for operational staff, management and the medical personnel. This problem can be solved using single objectives optimization and multi objectives optimization. In previous research, the scheduling optimization problem was solved single-objectively using Native Binary Particle Swarming Optimization (NBPSO). In this research, the researcher attempted to solve the problem multi-objectively using Non-Dominated Sorting Native Binary Particle Swarming Optimization (NSNBPSO). In this research, a constraint handling method was proposed called deallocation method which produced schedules that met the requirement of all constraints. In the performance comparison, researcher compared the performance of the proposed NSNBPSO and the previous NBPSO. The metrics used for the comparison were Coverage Rate, Hypervolume and Processing Time.NSNBPSO successfully produced 10 Solutions which provided more options to the stakeholder. However, NSNBPSO produced infeasible solutions and violated constraints in this research. Therefore, deallocation method was proposed and implemented to help NSNBPSO producing feasible solutions. The performance result showed that NSNBPSO with deallocation method had 35% more average hypervolume value than NBPSO, 50% of solutions from NSNBPSO dominated solution from NBPSO in coverage rate evaluation. However, it needed 85% more processing timeandcan still be considered acceptable.