See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/261499321

The Home Care Scheduling Problem: A modeling and solving issue

Conference Paper · April 2013

DOI: 10.1109/ICMSAO.2013.6552619

CITATIONS READS
11 1,083

3 authors, including:

Jaber Jemai Marouene Chaieb

Higher Colleges of Technology Institut Supérieur de Gestion de Tunis
9 PUBLICATIONS   85 CITATIONS    6 PUBLICATIONS   21 CITATIONS   

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Marouene Chaieb on 17 October 2015.

The user has requested enhancement of the downloaded file.

 The Home Care Scheduling Problem: A modeling
and solving issue

Jaber JEMAI Marouen CHAIEB Khaled MELLOULI

IS Department LARODEC Laboratory LARODEC Laboratory
CCIS - Imam University ISG - University of Tunis IHEC - University of Carthage
Riyadh - Kingdom of Saudi Arabia Tunis - Tunisia Tunis - Tunisia
Email: [email protected] Email: [email protected] Email: [email protected]

home. That is by employing a medical or a paramedical staff.

Intuitively, many questions need to be answered such as:
Abstract—The Home Care Scheduling Problem (HCSP) con-
sists of designing a set of routes for medical service suppliers to
visit a number of patients located at different places and asking 1) Which staff member (or team) will be assigned to
for a set of medical tasks. The object of the HCSP is to define the each patient?
required set of routes where the assigned workforce will travel
in order to minimize the overall travelled distance. Therefore,
2) Will nurses and caregivers will move alone or in
the solution will define first the teams (crew) that will travel groups to patients homes? If some grouping is re-
together for serving particular subset of patients. Then, the route quired, how to cluster them?
of each team is defined. The problem shows a hierarchical nature 3) What are the routes to be followed by staff groups in
where the solving algorithm should build first the service team order to move to patients locations?
that defines partially the set of potential customers to serve and
then look for the routes to follow. In this paper, we studied and
solved the HCSP after reviewing the literature on the HCSP. We It is clear that the answer of each question is not trivial and
modeled the HCSP as a multilevel (hierarchical) combinatorial each question represents an independent optimization problem.
optimization problem. Then, we focused on particular variant of The first question defines an assignment problem where the
the HCSP which is the Multiple Traveling Salesman Problem with objective is to design an optimal affectation (assignment) of
Time windows (MTSPTW) where we implemented a Tabu Search nurses to patients. The second question can define a clustering
based heuristic to solve a modified version of Solomons instances. (grouping) problem and the aim is there to define the groups
The obtained results are reported, commented and validated.
of caregivers who will move together using the same trans-
portation tool. The third question is a routing problem that
I. I NTRODUCTION defines a set of routes to be followed by medical groups to
visit patients at their homes. All of these problems are highly
Home or domiciliary care is the provision of continuous constrained. We can cite, staff workload constraints, routing
and coordinated cares (medical and paramedical services) to constraints like time windows, etc. Moreover, these questions
patients in their own homes, according to a formal assessment and problems are not independent, in the sense that the answer
of their needs with the aim of maintaining or improving their of the first question will be an input to the second question and
life conditions. This support is made by medical suppliers that then the answer of the second question will be the input of the
are generally nurses, drug vendors, technicians, etc. The aim third question. For instance, the proposed assignment (answer
is to provide the care and support needed to assist people. The of question 1) will affect the design of the nurse groups that
population taking benefit for home care scheduling contains, will move together. For example, nurses assigned to patients
mainly of elderly peoples and people with physical or learning in the same city will move together. Following the defined
disabilities. The Home Health Care (HHC) services are a groups, the routes will be designed in the third level. Thus,
growing service industry which must face scheduling and the home health care scheduling problem asks explicitly for
routing problems. Its development is accelerated by several answering patients demands for service at their homes. But,
factors such as the introduction of innovative technologies, implicitly, it is a sequence of three dependent optimization
the population ageing, the increase in chronic pathologies problems. Such optimization problems are called Hierarchical
and the continuous pressure of governments to contain health problems or dynamic problems. Hierarchical problems are NP-
care costs. From the stakeholder (hospitals, agencies) position, Hard Combinatorial Optimization Problems.
the objective of the HHC is to ensure the efficiency of the
service of patients. Service efficiency is defined by customers
The next section provides a review of the HCSP literature
satisfaction and also the cost incurred by the completion of the
by presenting its main components. Section 3 presents the
service.
proposed approach, describes the optimization model and
Home health care services can be viewed as the problem presents the solving method. In the fourth section, we describe
of satisfying a set of patients queries to be served at their the implementation and the computational results. Finally, we
discuss the conclusions and areas of future research of the
978-1-4673-5814-9/13/$31.00 ⃝2013
c IEEE HCSP in Section 5.
 II. L ITERATURE REVIEW focus on assignment in health care: Powers [37] suggests
that excessive workloads are unsafe and promote poor quality
The literature on the HCSP, shows three types of studies of care. Moreover, fair workloads among nurses are essen-
of the problem: in first class the focus is on the assignment tial for optimal quality of care as agreed by [19] [15] and
problem, in the second class of papers the main studied [23]. Walts et al. [45] develop a patient classification system
part is the routing problem and in some recent papers the and use an integer program to model and compute minimal
problem is handled without omitting or hiding one of its staffing levels required to meet aggregate workload levels.
two components. Subsequently, the next three subsections will Their approach provides significant reduction in staffing levels
present the relevant literature in each class. over current practice, while ensuring that nursing units are
adequately staffed. Shaha and Bush [40] say that most patient
A. HCSP: the assignment component assignments are based either on the intuitive judgment of the
The first part of the home health care scheduling problem charge nurse or on the caseload method wherein each nurse
consists of defining the assignments of nurses to patients. is assigned the same number of patients. Punnakitikashem et
The proposed assignment has to respect the type of the al. [38] mention some researches studied nurse assignment to
service demanded by the patient. Generally, each treatment patients like the work of Overfelt [35] in which he proposed
requires one or more caregiver skills. The visits have to be a modern patient classification systems partition the set of
conducted within particular periods that can be specified by patients into groups, and each group is assigned to a nurse.
the patients or by the type of the treatment itself. Moreover, Punnakitikashem et al. [38] developed a two stage stochastic
caregivers assignments will respect employment regulations integer programming model for a nurse patient assignment that
and contracts in term of nurses overall workload, breaks, etc. considered uncertainty in patient care. Sundaramoorthi et al.
In this context, Ben Bachouch [4] described some of the [41] presented a simulation model from real data to evaluate
considered constraints: nurse-patient assignments. Teoh [43] provided an overview
of the workload measurement systems and its application
• The biggest workload of each nurse must be lesser to balance workload between workers. In [28], authors aim
than the upper bound of the workload of each nurse to balance the workload of the nurses while avoiding long
Pmax . travels to visit patients. Abdennadher [1] discussed the nurse
scheduling problem and a specific system named INTERDIP,
• The smallest workload of each nurse must be greater is presented using constraint logic programming. This system
than the lower bound of the workload of each nurse assists a human planner in scheduling the nurse working shifts
Pmin . for a hospital ward and provides a feasible planning in few
• Each task is assigned to one and only one care worker minutes. Eiselt [21] in which they proposed a model for the
according to a given rank. assignment of tasks to individual employees, when several
goals are considered and when there are constraints imposed
• All the tasks must be realized into the working day of by employees capabilities.
the care workers.
• etc. B. HCSP: the routing component
In [28], Hertz and Lahrichi describe the following constraints: The goal of the HCSP routing problem is to find the
optimal set of routes to be followed by care providers vehicles
• The sum of work load of each nurse must be less than to reach patients home and to complete the required treatments.
or equal to the average work load of all nurses. To achieve a good scheduling, the routing component must
• The sum of patients assigned to a nurse must be less consider the already proposed assignment and minimize trans-
than or equal to the average of the number of patients portation costs. Routing costs consist generally of minimizing
affected to all nurses. the number of used vehicles, the overall travelled distance,
balancing tours in term of staff workload and the overall
Generally, the objective of the assignment task is to balance working time (accumulated durations of runs). The HCSP
the workload between workers [4], [28] in order to enhance routing problem is subject to a large number of scheduling and
quality for patient caring and increase the quality of nursing routing constraints. According to [22] and [10] , the optimal
works. Consistently balanced workloads help nursing man- solution requires the respect of the following constraints:
agers to predict required staffing levels and identify over-
staffed units more easily. Furthermore, distributing work fairly • Each nurse have to respect the proposed assignment
among nurses is essential for optimal quality of care. However, and he must visit the affected patients in a particular
other objectives may be considered like maximizing benefits order.
incurred by serving more patients with minimal number of • Each patient have to be visited exactly the number of
caregivers. Given the number of potential objectives, the as- required visits generally set to one visit.
signment problem can be modeled as single-objective or multi-
objective problem [28]. • Care providers visits must be carried within the pre-
defined time window.
Assignment in health care problem is a consistent part of
the assignment literature. Assigning in home health service • Each staff member has given planned breaks such as
is about the finding the best affectation of medical staff meals; sufficient travel time between visits must be
to patients. Mullinax and Lawley [34] cited some works allocated.
 • Staff members visits will complete their circuits start- In dynamic HCSP, both assignment and routing constraints
ing from the central point and then going back home are defined and have to be respected in an a final feasible
at a defined time. solution. Chananes et al. [16] and Bertels [8] presented some
constraints to guarantee both good assignment and routing
• Additional operational constraints may be found like tasks that we can cite:
the precedence constraints between visits. For in-
stance, a blood sampling visit has to precede any • Each activity must be delivered within a specified time
treatment (second visit) determined by the results of window and location.
the blood analysis.
• Each activity can involve only one visit by one care
Health care routing problem was extensively studied to worker so the maximum number of routes is equal to
model and solve the routing component of the HCSP. In [22], the number of care workers.
authors said that repeated matching was first used by Forbes • The care workers start from their homes and return
[25], in bus crew scheduling is the basis for solving vehicle after finishing all their assigned activities.
routing problems. There are many articles devoted to staff
scheduling. One area is the airline industry, many works in • For critical, medical activities the time window is a
this domain [29] and [26] showed that improved schedules target time of 5 min. For non-critical activities the
has a considerable positive impact. In some applications, the window is 15 min.
schedules must be cyclic and examples of such applications • If a care worker arrives before a time window, the
are found in [32] and [7]. Problems also differ whether the service cannot begin and a waiting- time is incurred.
definition of a shift or task. If the schedule is determined in If a care worker arrives after time window (late), the
advance [32] and [6]. One area in the service industry that has solution is infeasible.
attracted lot of attention is hospital and/or nurse scheduling
we have the example of [20][44]. Also in the area of call • The maximum capacity of each worker is 7.5 h per
centers like the work of Lin et al. [30]. Many heuristic methods day, including travel time.
have been used in the literature that we can mention the use
• Each worker is assumed to be available 24 h per day,
of genetic algorithms [13], simulated annealing [12] and tabu
but can be used for only 7.5 h in that period.
search [20]. A variety of optimization based methods constraint
programming, column generation, set partitioning models was • The travel speed of a care worker is assumed to be 30
also used [14][11][39]. In [8] and [10], authors developed a miles per hour and traffic conditions are ignored.
computerised system which solves the nurse scheduling and
the routing home care nurses problems. Cheng [18] formulated • All jobs must be covered at one and only time, any
the routing home care nurses problem as a vehicle routing starting point for a job has to respect the hard time
problem with time windows and multiple depots. Eveborn [22] window.
introduced a scheduling problem for a variety of home care
Many different and often conflicting objectives may be con-
providers which is modeled as a set partitioning problem and
sidered in the HCSP. Such objectives can be derived first
solved with a repeated matching algorithm. Cheng [18] cited
from the assignment problem objectives and also those of
some of works about routing where the goal was to minimize
the routing problem. We can find then the following dynamic
the amount of work by minimizing the amount of travel
HCSP potential objectives:
needed. They considered that the home health care problem
is, more generally, a vehicle routing problem with and without • To reduce the traveling distance and hence traveling
time windows (VRPTW). costs of the care workers.
• To improve worker utilization by reducing the waste
C. The hierarchical home care scheduling problem of travel and consequently reducing the number of
workers required.
The hierarchical home care scheduling problem is a dif-
ficult task that aims to optimize both assignment and rout- • To increase customer service by satisfying all service
ing tasks. The problem is to find a feasible working plan requirements within their specified time windows.
for all nurses that have to respect a variety of hard and • To satisfy employee preferences.
soft constraints, and preferences. It looks for an operational
working plan for the nurses without separating the assignment • To distribute work equally.
component and the routing component.
The HCSP was, basically, viewed as an assignment prob-
The HCSP can be viewed as multi-level optimization lem with a less important routing component or routing
problem called also hierarchical optimization problem where problem with some assignments. That is the focus was set
decisions are token at different levels and each decision will on the routing part or the assignment part. However, some
affect the optimality of its following problem. For instance, interesting works consider the problem as unit and gave the
the assignment obtained at the first level (the assignment same importance to the routing and the assignment and they
component) will define the level of optimality of the routing conclude that an efficient scheduling has to optimize while
problem. Then and in order to achieve a global optimal assigning and also while routing [16]. The authors in [16] cited
working plan, it is obvious to consider the different parts (level) some works in this field like [46] in the domain of airline crew
of the HCSP as a unit without partitioning it into sub-problems. scheduling, the work of Bialas et al. [9] in the domain of bus
and truck driver scheduling and [17] for optimizing a routing For some variants like the MTSPTW the objective is clear
problem in call centers. A decision support system called and the problem is well defined. But, for other instances the
LAPS CARE has been developed to aid the staff-planning major or primary objective is generally defined by decision
task [22]. Chananes [16] presents the novel application of a makers. They have to mention which problem or component
collaborative population-based meta-heuristic technique called (assignment or routing) is more important than the other.
Particle Swarm Optimization (PSO) to the scheduling of home Another more natural and interesting alternative is to model
care workers. Bertels [8] cited some works about dynamic the HCSP as a multiobjective optimization problems. In the
HCSP like in [18] that described a combined mixed-integer remaining of this project, we will focus on modelling and
programming (MIP) and heuristics approach. In [7] a decision solving the MTSPTW.
support system that is based on simple scheduling heuristics
was proposed. Most of these use constraint programming A. The Multiple Traveling Salesman Problem with Time Win-
techniques in order to model and solve the nurse rostering dows
problem. Classical vehicle routing with time windows [24][27]
reflects the mobility aspect of the problem, but ignores any Basically, the multiple traveling salesman problem with
further restriction. time windows is an extension of the canonical Traveling
Salesman Problem (TSP). The TSP asks for defining the
optimal route for a salesman to visit a set of customers
dispersed on a transportation network. In the basic TSP, a
III. HCSP MODELING AND SOLVING A PPROACH feasible solution must ensure that each customer will be
Health home are services can be analytically represented visited exactly one. The MTSPTW consists of defining a
by the HCSP as shown in the previous section. However, the TSP with Time Windows TSPTW for each medical staff and
HCSP can be viewed as general problem where all kinds of then design the route to follow for that caregiver. Additional
home health care services can be converted to. Besides that, constraints related to time windows and workload have to be
some real life HHC problems may be less restrictive than considered. The TSPTW has intended recently some interest
the basic HCSP depending on the hospital or agency business to model and represent particular optimization problems
rules. We can find many variants of the HCSP following the arising in real life and also to find lower bounds to evaluate
considered assumptions and hypothesis. In the following we approximate optimization [2][33]. In the following, we report
will describe some particular variants of the HCSP based on the Big-M formulation from [3]. The Big-M formulation
the number of different skills of a nurse and the number of defines the decision variable xij to be 1 if i is followed by
treatments required by a patient. j in the final solution or not. The variable si represents the
time where the service starts at customer i. Following is the
1) If the HCSP suppose that each nurse has one skill Big-M formulation where V is the set of vertices and ei and
and each patient will ask for one treatment, then li are respectively the earliest and the latest service time of
the assignment component is implicitly removed and customer i:
the problem will consist on routing the set of care- ∑
M in xij dij (1)
givers to their already defined patients. However, i,j∈V
more HCSP constraints are maintained. This kind of Subject to:
problems can be modeled as a Multiple Traveling ∑
Salesman Problem with Time Windows where a TSP xij = 1 ∀j ∈ V (2)
with time window is defined for each nurse. ∑
j∈V
xji = 1 ∀i ∈ V (3)
2) If each staff member is described with one skill and i∈V
patients may ask or more than one treatment (skill), si + tij − (1 − xij )Mij ≤ sj ∀(i, j) ∈ V (4)
then the routing problem is less important than the ei ≤ si ≤ li ∀i ∈ V (5)
assignment subproblem. The major focus here is on xij = 0 ∀i, j ∈ V (6)
which nurse will be assigned to serve each patient. where M ij = ei − lj + tij . The constraints (2) and (3)
Therefore, the objectives are generally those of the as- ensure that the xij variables with value 1 in a feasible solution.
signment component mainly workload balance. Such The constraints (4) ensure that start times at the nodes are
problems are generally assignment problems with a increasing along any path (as tij > 0) and therefore directed
post-assignment part to define the routes to follow. cycles cannot exist in the solution. Finally, the constraints
3) Similarly to the previous case, if each nurse come (4) and (5) together ensure that the solution respects time
with one skill and patients demand many treatments, windows.
the primary aim of the problem is to find the assign-
ment and then design the paths for reaching patients
B. Solving approach: Tabu search
home.
4) Suppose that nurses can provide many treatments and The Operations Research (OR) field offer a huge toolbox
the patients may ask for many nurses skills, then the of methods, techniques, algorithms and software for solving
problem will return to its initial definition and both combinatorial optimization problem. We can find exact algo-
parts (assignment and routing) will have the same rithms that find the optimal solution of the studied problems,
importance for optimizing and then finding the best but they fail where the problem increase in dimensionality.
operational decisions. For this particular type health Particularly, exact algorithm like branch and bound, dynamic
care variants, the problem have to be tackled without programming, etc are suitable for solving small problem. For
dividing it into two subproblems. NP-hard class of optimization problems, obtaining the optimal
 TABLE I. D ETAILED RESULT OF INSTANCE : C101
in a reasonable time is not trivial. That is, a class of solving
approaches were proposed to give a good compromise between Nurse No Initial distance final distance Time (ms)
the cost of the obtained solution and the time needed to find 1 461 258 537
it. They are called heuristics or generally metaheuristics [42]. 2 500 231 535
Metaheuristics represent a class of problem solving generic 3 512 297 589
algorithms generally inspired from nature. We can cite Genetic 4 520 254 643
Algorithms (GA), Simulated Annealing (SA), Ant Systems 5 554 257 624
(AS), Tabu Search (TS), etc. All these algorithms have proved 6 335 203 437
their efficiency, with variable performance, in solving and 7 543 284 742
handling a wide variety of optimization problems. Choosing 8 410 251 526
a particular metaheuristic for solving an optimization problem 9 484 235 531
depend generally on the number of parameters to set and if the 10 346 239 285
heuristic is convergent asymptotically to an optimal solution.
For solving the TSPTW, we select to implement a Tabu Search TABLE II. D ETAILED RESULT OF INSTANCE : R101
based heuristic. Our choice is explained by the reduced number
Nurse No Initial distance final distance Time (ms)
of the TS algorithm parameters and also the experimental
1 277 186 486
performance of the TS on routing problems [27].
2 402 213 766
3 308 205 397
IV. S IMULATION , I MPLEMENTATION AND 4 352 222 320
C OMPUTATIONAL RESULTS 5 296 178 333
A. Instances simulation 6 463 227 400
7 393 205 769
In order to simulate MTSPTW instances, we choose to 8 305 190 490
modify the 6 sets of Solomon’s benchmarks of the Vehicle 9 416 209 399
Routing Problem with Time Windows (VRPTW). The modi- 10 367 214 580
fications applied to standard Solomons benchmarks are:
TABLE III. D ETAILED RESULT OF INSTANCE : RC101
1) Vehicles are considered as nurses and to each nurse
is associated a skill si . Nurse No Initial distance final distance Time (ms)
2) Customers are transformed to be patients and each 1 563 301 550
patient asks for a particular treatment represented by 2 468 267 463
the required skill sj . 3 424 279 382
3) Time windows, locations ar maintained. 4 404 236 458
5 543 304 552
B. Experimental context and parameter settings 6 457 295 468
7 365 277 461
The HCSP solving approach has all been implemented 8 426 293 383
using the tabu search metaheuristic from the ParadisEO- 9 416 214 460
MO library [36]. ParadisEO is a C++ LGPL open source 10 433 237 463
object-oriented framework for evolutionary computation that
has been developed through an European joint work. The
rich set of ParadisEO-MO modular classes are combined to
V. C ONCLUSIONS
develop single solution based metaheuristics. The ParadisEO-
MO module is based on a clear conceptual separation of the The interest in the problems of health is growing in the
solution methods from the problems they are intended to solve. current life. Many areas are involved such as home care
This separation confers a maximum code and design reuse to services and emergency services. In addition, requirements
the user. The results presented below are base on the following of patients and environmental issues provide a significant
Tabu search parameters: importance to health costs. The HCSP is is by definition a
hierarchical optimization problem where the assignment and
• Initial solution is randomly generated.
routing components should be solved mutually and depen-
• Tabu list size=100. dently. However, the large part of HCSP contributions address
on of the components and neglect another. In this paper, we
• Neighborhood generation operators are the 2-opt and review present HCSP literature and we propose a modeling
Swap operators. and solving issue. We announced the assumptions under which
• Stopping condition: maximum number of iterations is the HCSP can be reduced to a TSP with Time windows.
1000. Consequently, we proposed and implement a tabu search
based heuristic for solving modified version of the Solomon’s
C. Computational results benchmarks of the VRPTW.
To demonstrate the efficiency of the implemented TS
heuristic for solving the modified solomons benchmarks, we R EFERENCES
report the following results in the next Table for the the [1] Abdennadher S. (1999), Schlenker H. Nurse scheduling using constraint
geographical data. logic programming. In: Proceedings of the 11th conference on innovative
 applications of artificial intelligence. Menlo Park, CA: AAAI Press; p: [26] Gamache M., Soumis F., Marquis G., Desrosiers J.(1999), A column
83843. generation approach for large-scale aircrew rostering problems, Opera-
[2] Albiach J., Sanchis J. M. and Soler D. (2008), An asymmetric TSP with tions Research 47,p: 247263.
timewindows and with time-dependent travel times and costs: An exact [27] Gendreau M. Metaheuristics in Vehicle Routing. ICORES 2012.
solution through a graph transformation, European Journal of Operational [28] Hertz A. and Lahrichi N. (2006), A patient assignment algorithm for
Research, Volume 189, Issue 3, p: 789802. home care service, les cahiers du GERAD.
[3] Ascheuer N., Fischetti M., Graotschel M.(2001), Solving the asymmetric [29] Hoffman K.L., Padberg M.(1993), Solving airline crew scheduling
travelling salesman problem with time windows by branch-and-cut, problems by branch-and-cut, Management Science 39, p: 657-682.
Mathematical Programming, Ser. A 90, p:475 - 506.
[30] Lin, C. K. Y., Lai, K. F., and Hung, S. L. (2000). Development of a
[4] Ben Bachouch R. (2010) An optimization model for task assignment workforce management system for a customer hotline service. Computers
in home health care. IEEE Workshop on Helath Care Management and Operations Research, 27, p:9871004.
(WHCM10), Venise: Italie. [31] Mason A.J., Ryan D.M., Panton D.M.(1998), Integrated simulation,
[5] C. Barnhart, E.L. Johnson, G.L. Nemhauser, M.W.P. Savelsbergh, P.H. heuristic and optimisation approaches to staff scheduling, Operations
Vance (1998), Branch-and-Price: Column generation for solving huge Research 46, p: 161175.
integer programs, Operations Research 46 , p: 316332. [32] Mason A.J., Smith M.C. (1998) A nested column generator for solving
[6] Beasley J.E., Cao B. (1996), A tree search algorithm for the crew rostering problems with integer programming. In: Caccetta L, Teo KL,
scheduling problem, European Journal of Operational Research 94, Siew PF, Leung YH, Jennings LS, Rehbock V, editors. International
p:517526. conference on optimisation: techniques and applications. p: 827834.
[7] Beaumont N.(1997), Scheduling staff using mixed integer programming, [33] Mitrovic-Minic S., Krishnamurti R. (2006), The multiple TSP with
European Journal of Operational Research 98, p: 473484. timewindows: vehicle bounds based on precedence graphs, Operations
[8] Bertels S. (2006), Fahle T. A Hybrid Setup for a Hybrid Scenario: Research Letters, Volume 34, Issue 1, Pages 111120
Combining Heuristics for the Home Health Care Problem.Computers [34] Mullinax C. and Lawley M. (2002). Assigning patients to nurses in
and Operations Research; vol 33; p:2866-2890. neonatal intensive Care Journal of the Operational Research Society 53,
[9] Bialas W.F., Karwan M.H.(1982), On two-level optimization. IEEE p:25-35.
Trans. Automat. Control, AC-27, p: 211214 [35] Overfelt, F. (2004) Nursing ratiosare they good or bad? Proceedings of
[10] Bouazza E., Jacques A. F and Viviane G (2008). Mathematical Pro- the 16th Annual Society for Health Systems Management Engineering
gramming Approach for Routing Home Care Nurses, Proceedings of the Forum, Orlando, Florida.
IEEE IEEM. [36] Paradiseo home page: http://paradiseo.gforge.inria.fr/, Last visit:
[11] Brailsford S.C., Potts C.N., Smith B.M.(1999), Constraint satisfaction 05/06/2012
problems: Algorithms and applications, European Journal of Operational [37] Powers J. (1993), Accepting and refusing assignments. Nursing Mngt
Research 119, p:557-581. 24, p:64-73.
[12] Brusco M.J., Jacobs L.W.(1993), A simulated annealing approach to [38] Punnakitikashem P., Rosenberger J.M., and Behan, D.B. (2005)
the cyclic staff-scheduling problem, Naval Research Logistics 40 6984. Stochastic programming for nurse assignment. Technical Report COS-
[13] Cai X., Li K.N.(2000), A genetic algorithm for scheduling staff of mixed MOS 05-01, The University of Texas at Arlington.
skills under multi-criteria, European Journal of Operational Research 125 [39] Ribeiro C.C., Soumis F.(1994), A column generation approach to the
359369. multiple-depot vehicle scheduling problem, Operations Research 42,p:
41-52.
[14] Caprara A., Focacci F., Lamma E., Mello P., Milano M., Toth P.,
Vigo D. (1998), Integrating constraint logic programming and operations [40] Shaha S. and Bush C. (1996). Fixing acuity: a professional approach
research techniques for the crew rostering problem, SoftwarePractice and to patient classification and staffing. Nursing Econ 14: p:346-356.
Experiences 28, p:4976. [41] Sundaramoorthi, D., Chen, V.C.P., Rosenberger, J.M., Kim, S.B., and
[15] Catterson J. (1988). How busy are you? Nursing Times 11, p: 2831. Behan, D.B. (2006) A data-integrated simulation model to evaluate nurse-
patient assignments. Technical Report IMSE 06 - 05, The University of
[16] Chananes A., Pisal Y., Paul R. D.(2007), PSO-based algorithm for home
Texas at Arlington.
care worker scheduling in the UK. Computers and Industrial Engineering
53, p: 559583. [42] Talbi E-G., Metaheuristics: from design to implementation, Wiley, 2009.
[17] Cheang, B., Li, H., Lim, A., and Rodrigues, B. (2003). Nurse scheduling [43] Teoh Sheu Siew C. and Abdul Ghani N. (2006), An overview of nurses
problems A bibliographic survey. European Journal of Operational workload measurement system and workload balance, proceedings of
Research, 151, p:447460. the 2nd IMT-GT Regional Conference on Mathematics, Statistics and
applications, Universiti Sains Malaysia, Penang, June 13-15, 2006.
[18] Cheng E., Rich J.L. (1998), A Home Health Care Routing and Schedul-
ing Problem. Technical report CAAM TR-04. Rice University. [44] Valouxis C., Houses E. (2000), Hybrid optimization techniques for the
work-shift and rest assignment of nursing personnel, Artificial Intelli-
[19] Drennan V. (1990). Striving for fairer workloads. Nursing Times 10, gence in Medicine 20, p:155-175.
p:12-14.
[45] Walts L. and Kapadia A. (1996). Patient classification system: an
[20] Dowsland K.A. (1998), Nurse scheduling with tabu search and strategic optimization approach. Healthcare Manager Rev 21: 75-82.
oscillation, European Journal of Operational Research 106, p:393-407.
[46] Yan, S., and Tu, Y. (2002). A network model for airline cabin crew
[21] Eiselt H. A. and Mirianov V. (2008), Employee positioning and work- scheduling: Case study. European Journal of Operational Research, 140,
load allocation, computers and operations research, vol 35, p: 513-524. p:531-540.
[22] Everborn, P., Flisberg, P., and Ronnqvist, M. (2006). LAPS CARE An
operational system for staff planning of home care. European Journal of
Operational Research, 171, p:962-976.
[23] Farnham J.A., Maez-Rauzi V. and Conway K. (1992). Balancing
assignments: a patient classification system for a step-down unit. Nursing
Mngt 23, p: 49-54.
[24] Fisher ML.Vehicle routing. In: Ball MO, Magnanti TL, Monma CL,
Nemhauser GL, editors. Network routing. Handbooks on operations
research and management science, vol. 8, Amsterdam: North-Holland;
1995. p: 1-33.
[25] Forbes M.A., Holt J.N., Kilby P.J., Watts A.M. (1991), A matching
algorithm with application to bus operations, Australasian Journal of
Combinatorics 4, p: 7185.

View publication stats