A room-oriented artificial bee colony algorithm for optimizing the patient admission scheduling problem
Introduction
The ultimate goal of a hospital care provider is to provide an efficient and high-quality service to its clients. However, the frequent surge in the population of clients seeking such services is thus making the goal increasingly more difficult to attain without some specific automated systems, especially in managing its limited resources among the various entities and events competing for such services. One area in the hospital where such automation is crucial is in the management of patients seeking admission into a hospital. The patient admission scheduling (PAS) problem, as it is usually called, is one of the challenging scheduling problems faced by many medical institutions that deal with finding an optimal way of assigning the hospital resources (such as beds, rooms, timeslots, and so on) to patients seeking the hospital admission while meeting all the necessary hospital requirements. Such hospital requirements or policy may differ from one hospital to another but could include that patients in a room must be of the same gender, accommodating patients could make a special room request, etc. Formally, PAS as a class NP-hard problem that belongs to the category of combinatorial scheduling problem entails the assignment of limited resources like rooms, beds, specialism, and timeslots to a number of patients so as to satisfy a set of restrictions [1]. Note that in a PAS problem, the number of given constraints are typically categorized into soft and hard. The constraint which needed to be satisfied for the PAS solution to be feasible is called the hard constraints while the violation of soft constraint is allowed in a feasible solution to the PAS. However, the reductions of such violations should be carried out to the barest minimum. The quality of the solution to the PAS is measured by the violations of soft constraints in a feasible PAS solution and the objective is to produce a high-quality solution to the PAS problem that satisfies the required constraints (i.e feasible solution). It is worthy of mentioning that studies have proven that it is almost impossible to optimize a feasible solution to the PAS to the tune of satisfying all the soft constraints [2]. Different formulations of the PAS have been proposed and investigated over the last few decades. Few examples of the literature that worked on formulations of other hospital versions can be studied in [3], [4], [5]. Similarly, the most widely investigated formulation of the PAS problem developed in [6] and investigated by [7] is the focus of this study.
The patient admission scheduling problem (PAS) has drawn the attention of research communities in the field of timetabling, scheduling, and operational research over the last few years in which numerous studies have proposed several methodologies to tackle the PAS problem. Several artificial intelligence approaches introduced over the years are traditional-based methods, metaheuristic-based methods, hyper-heuristic methods, and hybrid-based methods. Some examples of commonly used traditional methods are column generation-based approach [8], [9], mixed-integer programming [10], [11] and linear programming-based approach [3], while those employed under metaheuristic-based are classified into two: local search-based and population-based methods. Few examples of the local search-based algorithm are adaptive non-linear great deluge algorithm [12] and simulated annealing (SA) [1], [13], [14]. Furthermore, the techniques that made used of population-based include genetic algorithm (GA) [4], [5] biogeographical-based optimization algorithm [15], [16], [17], and harmony search algorithm [18]. Similarly, the usage of hyper-heuristic techniques for the PAS problem has been reported in [7], [19] and lastly, the hybrid-based approach utilized to tackle the PAS problem can be found in [20]. Other studies on patient admission scheduling problems can be found in [21], [22], [23], [24].
A critical analysis of existing techniques that worked on the instances of the PAS under consideration from the literature revealed that the results obtained by the population-based metaheuristics like BBO [15], [16], [17] and HSA algorithms [18] are not as good as those obtained by other techniques. This is due to the fact that these algorithms tend towards the exploration of the solution space rather than exploitation and thus lose the capability of exploiting the rugged nature of the search space of the PAS. In view of the above, the motivation of this study is to introduce a new population-based swarm intelligence algorithm with the incorporation of efficient neighbourhood structures for tackling the PAS problem.
In 2005, Karaboga introduced Artificial Bee Colony (ABC) algorithm as a population-based metaheuristic to solve practical optimization problems [25]. It is relatively a new nature-inspired optimization algorithm that simulates the intelligent foraging behaviour of honey bees in their hive. Experimentally, ABC has the advantage of utilizing fewer control parameters, it has a competitive performance in comparison with other swarm intelligence-based techniques. The simplicity and ease of implementation of the ABC motivated its very high utilization in solving many practical combinatorial optimization problems [26]. Few examples of the complex optimization problems in which the success of ABC have been recorded are timetabling problems [27], [28], [29], Job shop scheduling problem [30], Vehicle routing problem [31], travelling salesman problem [32], protein structure prediction [33], and well placements in fractured reservoirs [34]. Note that other notable areas of application of the ABC algorithm can be found in [26], [35]. Based on the competitive performance of the ABC algorithm, therefore, the main contribution of this paper is to adapt the ABC algorithm with the integration of efficient and simple neighbourhood searches to its operators as an alternative solution template to the research community when utilized to tackle the PAS problem. The performance of the proposed ABC algorithm for the PAS problem is evaluated using the popular Benchmark dataset of the PAS consists of 13 real-world problem instances as proposed in [6] and investigated in [7]. It is noteworthy that the proposed ABC algorithm is tested on the six most widely utilized instances from this benchmark dataset. The results of the experiment produced by the proposed ABC method achieved the first rank when compared with population-based and other heuristic methods and came third in comparison with hybrid integer programming methods
The outline of this paper is presented as follows: the descriptions and formulations of the PAS are provided in Section 2 while Section 3 discusses an overview of the ABC algorithm. Section 4 describes the adaptation of the proposed ABC algorithm for the PAS problem and presentation of the experimental analysis, discussion of results, and comparison with other state-of-the-art techniques are discussed in Section 5. Lastly, the conclusion and future directions are presented in Section 6.
Section snippets
Descriptions of the problem
The problem to the PAS could be described based on some basic features that include the patients, rooms, wards, and timeslots in which elements like treatment demands (i.e. nursing and medical equipment), age category, room preference, gender, admission date, and the length of stay defines a patient. Room attribute is given as the number of beds, the ward in which the bed is located, and medical equipment. Note that the treatment a ward could offer is based on the types of qualities of the
Artificial Bee Colony Algorithm
ABC algorithm is a new stochastic algorithm that utilizes the principle of natural phenomenon in solving the non-deterministic problem. Basically, ABC algorithm is a population-based nature-inspired technique that belongs to the class of swarm intelligence based algorithms, where the behaviours of a group of bees are studied based on the idea that mimics the intelligent foraging behaviour of honey bee in their colony. For instance, a swarm is a collection of a set of honey bees that socially
Application of ABC for the PAS
In this section, the steps involved in solving the PAS problem using the ABC algorithm is presented. Similarly, three neighbourhood operators are incorporated with the components of the ABC algorithm in order to cope with the ruggedness of the PAS solution space. It is worthy to mention that the proposed ABC algorithm for solving the PAS maintains the feasibility of the solution space during the search process. The procedure of generating a feasible solution for the PAS problem is presented in
Computational experiments, results and discussions
This section presents the performance of the proposed ABC algorithm for the PAS problem. The proposed algorithm is coded using the edition of the JAVA NetBeans 6.0 under the window 7 platform. The experiments ran on an Intel Core(TM) i3-4005u CPU @1.70 GHz machine and 4 GB RAM. The performance of the proposed method is investigated on six benchmark instances of the PAS formulated by [6] which varied in sizes and complexity. The dataset is made available publicly and the features of the dataset
Conclusion and future research directions
This paper developed a swarm intelligence population-based metaheuristic for the PAS problem. The method framework utilizes the room-oriented approach to generate initial feasible solutions while the optimization of the solutions was carried out with ABC algorithm with the integration of efficient three neighbourhood structures to cope rugged nature of the PAS solution space. The main purpose of utilizing the ABC algorithm to solve the PAS problem lies in its ability to balance the exploration
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
References (40)
- et al.
Local search and lower bounds for the patient admission scheduling problem
Comput. Oper. Res.
(2011) - et al.
An optimization based on simulation approach to the patient admission scheduling problem using a linear programing algorithm
J. Biomed. Inform.
(2014) - et al.
An evolutionary approach to rehabilitation patient scheduling: A case study
European J. Oper. Res.
(2008) - et al.
A column generation approach for solving the patient admission scheduling problem
European J. Oper. Res.
(2014) - et al.
Mixed integer programming based heuristics for the patient admission scheduling problem
Comput. Oper. Res.
(2017) - et al.
A mixed integer programming approach to the patient admission scheduling problem
European J. Oper. Res.
(2019) - et al.
An adaptive non-linear great deluge algorithm for the patient-admission problem
Inform. Sci.
(2015) - et al.
An adaptive large neighborhood search procedure applied to the dynamic patient admission scheduling problem
Artif. Intell. Med.
(2016) - et al.
A hybrid tabu search algorithm for automatically assigning patients to beds
Artif. Intell. Med.
(2010) - et al.
University course timetabling using hybridized artificial bee colony with hill climbing optimizer
J. Comput. Sci.
(2014)