Tailoring a Red Deer Algorithm for Planning and

Scheduling Elective surgeries

Mohamed Tayeb Samed 1, Asma Ouled Bedhief2
1
Department of Industrial Engineering,
National Engineering School of Bizerte, University of Carthage
Tunis, Tunisia
e-mail address:
2
LR-OASIS, Department of Industrial Engineering,
National Engineering School of Tunis, University of Tunis El Manar,
Tunis, Tunisia
Department of Industrial Engineering,
National Engineering School of Bizerte, University of Carthage
Tunis, Tunisia
e-mail address: [email protected]

Abstract: This paper investigates an integrated elective surgery-planning and scheduling problem
considering simultaneously the operating rooms and the recovery beds. The objective is to minimize
the maximum daily closing time of the operating theatre. To solve this challenging NP-hard problem,
we developed a Red Deer Algorithm (RDA). To assess its performance, we have conducted numerical
experiments on a set of benchmark instances. Computational experiments demonstrate the
effectiveness of the proposed RDA in generating efficient surgical schedules, leading to enhanced
surgical efficiency and better patient outcomes.

Keywords: Operating rooms, Recovery Beds, Planning, Scheduling, RDA.

1 Introduction

Healthcare systems have been increasingly challenged in recent years to deliver high-quality care with
limited resources. Among the various departments within a hospital, the operating theatre (OT) stands out as
one of the most critical and expensive, as it comprises several key resources including operating and
recovery rooms and plays a pivotal role in hospital operations.
The operating rooms (OR) department, in particular, holds immense significance, accounting for 60% to
70% of all hospital admissions, primarily due to surgical interventions. Moreover, it is estimated that the OR
accounts for more than 40%of a hospital’s total expenses. These statistics highlight the necessity of efficient
management of the surgical process, which involves careful planning, scheduling, and resource coordination.
In this context, the integration of surgery planning and scheduling tasks has attracted increased attention in
recent literature. Surgery planning focuses on assigning a surgery date to each patient, while scheduling
involves allocating patients to resources and determining the order of surgeries on the scheduled day.
Integrating these tasks aims to improve coordination and optimize the performance of the surgical unit,
ultimately enhancing overall hospital efficiency.
Several studies have explicitly addressed the integration of planning and scheduling tasks. For instance,
Guido et al. [1] developed a multi-objective integer linear programming model aimed at optimizing hospital
operating room planning and scheduling. They also proposed a hybrid genetic algorithm approach to improve
convergence towards Pareto-optimal solutions. Similarly, Aringhieri et al. [2] introduced a tabu search
algorithm to solve an integrated OR planning and scheduling problem that considers post-surgery bed
availability. Their approach begins with a greedy constructive method to generate an initial solution, which is
then refined through a search strategy aimed at optimizing patient and hospital utilities. Extensive
computational experiments on real data sets demonstrated the effectiveness of the proposed metaheuristic.
Akbarzadeh et al. [3] tackled the integration of surgical case planning and scheduling by incorporating nurse
re-rostering and patient-nurse assignments, with the goal of maximizing operating room utilization and
profitability. They developed a heuristic approach based on linear programming and column generation to

1
 produce high-quality feasible solutions. Additionally, Doulabi et al. [4] proposed a branch-and-price-and-cut
algorithm for integrated OR planning and scheduling, focusing on the allocation and sequencing of surgeries
over a planning horizon. Their results indicated an average optimality gap of 2.81%. Wang et al. [5] addressed
planning and scheduling for laminar-flow operating rooms, which involves assigning surgeries to specific
dates while accounting for various resource constraints and minimizing overtime and undertime costs. To solve
this problem, they developed a particle swarm optimization algorithm combined with heuristic rules. Other
studies have proposed alternative solution methods for integrated surgery planning and scheduling problems
[6, 7, 8, 9]. For a detailed literature review, readers can refer to [10].
The integrated surgery planning and scheduling problem has received relatively little attention in the
literature, particularly when both operating rooms and recovery beds are considered simultaneously. This is in
contrast to the more extensively studied problems of surgery scheduling or planning individually. Most
research tends to focus either on assigning surgery dates (planning) or on sequencing and allocating patients to
resources (scheduling) as separate tasks. Additionally, to the best of our knowledge, no study has reported the
use of the Red Deer Algorithm (RDA) for solving the integrated surgery planning and scheduling problem.
This presents a notable gap in literature, as RDA could potentially offer an innovative and effective solution
for optimizing both planning and scheduling within a unified framework.
In this paper, we tackle an NP-hard integrated elective surgery planning and scheduling problem that
simultaneously accounts for both operating rooms and recovery beds. By considering these two critical
resources together, our goal is to develop a more comprehensive approach to surgical scheduling and resource
allocation.
Given a set of elective surgeries (patients) to be performed and a limited number of available operating
rooms and recovery beds, the challenge is to determine, within a given planning horizon: (1) the surgery date
for each patient; (2) the assignment of operating rooms and recovery beds to each surgery; and (3) the
sequence of surgeries in each operating room, recovery bed, and operating day. The objective is to create an
optimal schedule that minimizes the maximum daily closing time of the operating theatre, essentially
minimizing the makespan across all surgeries, operating rooms, and recovery beds.
In the surgery planning and scheduling problem under consideration, we make the following assumptions:
 A set of identical, multifunctional operating rooms is available during designated opening and closing
hours.
 Human and equipment resources are readily accessible whenever required.
 The duration of each surgery encompasses both the time needed to prepare the patient in the operating
room (e.g., anesthesia) and the actual surgical procedure.
 We adopt an open scheduling strategy, meaning that no restrictions based on medical specialties are
imposed.

The remainder of the paper is organized as follows: Section 2 presents a comprehensive description of the
Red Deer Algorithm (RDA) developed to address the integrated surgery planning and scheduling problem,
considering, simultaneously, the operating rooms and the recovery beds. In Section 3, we provide preliminary
computational results to assess the effectiveness of the proposed metaheuristic. Finally, the paper concludes
with a summary of findings and potential future research directions in the conclusion section.

2 Proposed Red Deer Algorithm

In this section, we describe the main phases of the Red Deer Algorithm (RDA) that we have adapted to address
the operating room scheduling and sequencing problem. Our focus is specifically on optimizing the use of
operating rooms and recovery beds.

a. Generating an Initial Solution

In this work, we propose that a solution to the problem be represented as a four-row array:

 The first row of the array represents the order of patients, each identified by a number from {1, 2, ...,
number of patients} This order is generated randomly.

 The second row of the array corresponds to the dates of surgeries assigned to patients, each date
represented by a number from {1, 2, ..., number of days in the planning horizon} Assigning dates to
patients is done randomly.
2
  The third row indicates the recovery beds assigned to patients, each bed designated by a number from
{1, 2, ..., number of recovery beds} Assigning recovery beds to patients is done randomly.

 The fourth row of the array indicates the operating rooms assigned to patients, each room designated by
a number from {1, 2, ..., number of operating rooms} Assigning operating rooms to patients is done
randomly.

For example, let’s consider a scheduling and sequencing problem with 10 surgical procedures for 10 patients.
We assume that only three operating rooms and two recovery beds are available within a three-day planning
horizon. A solution to the problem could be illustrated in Table III.1. Based on the surgical and recovery
durations, we can calculate the makespan associated with this solution. Our objective in this project is to find the
best possible solution that minimizes the makespan.

b. Generating an Initial Population

From the initial solution that we have generated, we construct a set of solutions where each solution is
associated with a makespan value (objective function) that reflects the solution’s performance. Thus,
we seek to optimize this value.

c. Red Deer Males Roaring

Males roar to enhance their prowess and attract females. Some succeed and become more successful
while others fail.

d. Selection of Commanders Among the Male Population

The most successful males succeed in attracting females and forming harems; they are thus called
commanders. The others are the weaker males known as Stags. The number of commanders is
calculated as follows:

N Commanders =Round( y × N Male )

3 Preliminary Results

The simulation results of the RDA algorithm, as well as the results obtained by the Linear
Programming Model (LPM), are presented in the table below.

The C max represents the makespan generated by the proposed Red Deer Algorithm (RDA) for the
corresponding problem instance. The C max indicates the optimal solution obtained by solving the
Linear Programming Model (MLP) that models this problem.

The makespan values generated by the RDA algorithm are very close to the optimal solutions
provided by the MLP, with a deviation not exceeding one hour. Thus, the solution obtained by the
RDA algorithm approaches optimality. It is important to highlight that the time required to solve
instances by the RDA algorithm is negligible, taking only a few seconds, compared to the time
required by the MLP. Although the resolution time for Cplex or RDA increases as the number of
patients increases, the time for RDA always remains less than that for Cplex.Tapez une équation ici .

Instance Patients Days Operating Recovery ( RDA )

C max Elapsed C max Elapsed Difference
s rooms beds time time in C max
(seconds (seconds)
)
A 10 5 2 2 3.75 3.9 3.75 76 0
3
 B 10 5 3 2 8 3.26 8 40 0
C 15 5 2 2 5.25 4.5 4.5 117 0.75
D 15 5 3 3 4.75 5.08 4.5 119 0.25
E 20 5 4 4 2 6.65 1..5 180 0.5
F 25 5 4 3 3 12.19 2 1200 1

4 Conclusion and Perspectives

The RDA algorithm, developed recently in 2020, has demonstrated its effectiveness by providing
satisfactory, and even superior, results compared to other heuristic and metaheuristic algorithms.
Through this approach, we have successfully minimized the makespan value to approach optimality.
However, the field of operating room management remains complex, with numerous constraints and
challenges. In our future steps, we plan to integrate additional constraints into the studied problem to
effectively and swiftly address real-world uncertainties.

References

[1] Guido, R., Domenico, C. (2017). A hybrid genetic approach for solving an integrated multi-objective
operating room planning and scheduling problem. Computers & Operations Research.
[2] Aringhieri, R; Landa, P; Soriano, P; Tànfani, Ea Testi, A (2015). A two-level metaheuristic for the operating
room scheduling and assignment problem. Computers & Operations Research, 54(), 21–34.
[3] Akbarzadeh, B., Moslehi, G., Reisi-Nafchi, M., & Maenhout, B. (2020). A diving heuristic for planning and
scheduling surgical cases in the operating room department with nurse re-rostering. Journal of Scheduling.
[4] Doulabi, S. H. H., Rousseau, L. M., & Pesant, G. (2016). A constraint programming-based branch-and-price
and-cut approach for operating room planning and scheduling. INFORMS Journal on Computing, 28(3), 432–
448.
[5] Wang, Yu; Tang, Jiafu; Pan, Zhendong; Yan, Chongjun (2015). Particle swarm optimization-based planning
and scheduling for a laminar-flow operating room with downstream resources. Soft Computing, 19(10), 2913–
2926. doi:10.1007/s00500-014- 1453-z
[6] Nyman, Jacob; Ripon, Kazi Shah Nawaz (2018). [IEEE 2018 IEEE Congress on Evolutionary Computation
(CEC) - Rio de Janeiro (2018.7.8-2018.7.13)] 2018 IEEE Congress on Evolutionary Computation (CEC)
Metaheuristics for the Multiobjective Surgery Admission Planning Problem. , (), 1–8.
[7] Di Martinelly, C.; Baptiste, P.; Maknoon, M.Y. (2014). An assessment of the integration of nurse timetable
changes with operating room planning and scheduling. International Journal of Production Research, 52(24),
7239–7250.
[8] Marques, Inês; Captivo, M. Eugénia; Vaz Pato, Margarida (2015). A bicriteria heuristic for an elective
surgery scheduling problem. Health Care Management Science, 18(3), 251–266.
[9] Atle Riise; Edmund K. Burke (2011). Local search for the surgery admission planning problem, 17(4), 389–
414. doi:10.1007/s10732-010-9139-x
[10] Cardeon B, Demeulemeester E, Belien J (2010). Operating room planning and scheduling: a literature
review. Eur J Oper Res 201(3): 921-932