Abstract

This article investigates a novel physician scheduling problem. Different types of tasks can be performed by physicians with certain qualifications. Tasks have different properties depending on their types, lengths, and starting times. Physicians performing tasks can yield different values of benefit and cost according to their qualifications and the task property. The objective is to maximise the sum of profit (i.e., benefit minus cost). For solving the studied problem, three layer-progressive processes are proposed and corresponding solution strategies are developed for them respectively. A Variable Neighbourhood Search is applied in the first-layer process to assign a certain qualification of physicians to each task property. The problem is then decomposed into scheduling physicians of single qualification as the second-layer process. On this layer, a heuristic incorporating a Dynamic Programming algorithm is developed to generate a task property list for each qualification of physicians to guarantee the optimum of the solutions. The Dynamic Programming algorithm is applied on the third-layer process to get the task property list for a physician. In the computational experiments, the proposed approach is compared with three meta-heuristic algorithms and Gurobi. The results show that the proposed approach outperforms other compared algorithms.

摘要

本文研究了一个新的医生调度问题。具有一定资格的医生可以执行不同类型的任务。任务根据其类型、长度和开始时间而具有不同的属性。执行任务的医生可以根据他们的资格和任务属性产生不同的收益和成本值。目标是最大化利润总和（即收益减去成本）。为了解决所研究的问题，提出了三层递进过程，并分别制定了相应的解决策略。在第一层过程中应用变量邻域搜索，为每个任务属性分配一定的医师资格。然后将问题分解为将单一资格的医生安排为第二层过程。在这一层上，开发了一种包含动态规划算法的启发式算法，为每个医师资格生成一个任务属性列表，以保证解决方案的最优性。动态规划算法应用于第三层流程，以获取医生的任务属性列表。在计算实验中，将所提出的方法与三种元启发式算法和 Gurobi 进行了比较。结果表明，所提出的方法优于其他比较算法。所提出的方法与三种元启发式算法和 Gurobi 进行了比较。结果表明，所提出的方法优于其他比较算法。所提出的方法与三种元启发式算法和 Gurobi 进行了比较。结果表明，所提出的方法优于其他比较算法。