การพัฒนาการจัดตารางรักษาแบบหลายวัตถุประสงค์ สำหรับผู้ป่วยปากแหว่งเพดานโหว่

Abstract

Nowadays, many hospitals usually face the treatment scheduling problem due to increasing the number of patients, resource limitations, and multiple procedures in the system. Also, the cleft lip and cleft palate patient treatment in Thailand is facing with scheduling problem that related to the complicated treatment and long treatment procedure. This research aims to simulate this problem and apply the optimization method to solve the multi-objective problem. The objectives of the patient's treatment scheduling are to minimize the total assigned hospitals' distance, minimizing the total assigned hospital preference value, and minimizing the total makespan. All objective values are formulated as a single objective function via weight-normalization method. In this research, the mathematical model was developed from previous research and then meta-heuristic methods namely Particle Swarm Optimization (PSO) was proposed with particular encoding and decoding schemes for solving the patient's treatment scheduling problem. The solution was tested on several scale numerical examples. To verify and validate the proposed mathematical model and algorithm, the small-scale instance was employed to primally test. The results showed that the patients' appointment arrangement was correct according to the scheduling constraint in this problem. After all scale instances were tested. The results showed that the exact method could excellently find the global solutions in the small cases but unable to solve larger-scale instances within the allowed computation run time. On the other hand, the heuristic method was able to find a feasible solution with computation runtime at most for 4 minutes. Besides, this research proposed other heuristic methods, including global and local neighborhood based PSO (GLNPSO) and Differential Evolution algorithm (DE) to solve the generated numerical examples and compare the evaluating result. The comparison results showed that PSO mostly was able to find a higher quality solution in the large-scale instance, while DE used the shortest run time to evaluate the feasible solution. Finally, this research is very useful to apply in real-life situations which makes sense for deciding on the best appointment date for the patient. From the obtained results within a short time will be convenient to plan work by simplifies the process of the patient appointment system, reducing the wasted time from finding a suitable appointment date, and maybe also lessening the congestion situation in the hospital. The result of the appointment scheduling in this research gives patients the convenience of being appointed to the nearest hospital. The ongoing research has continued to investigate ways to improve the algorithm performance for a wider range of patient's treatment scheduling problems.