Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Assignment 1

1CM190 Health care operations planning

Student name, number:


P.G.H. (Pien) Levels, 1334735

Nico Dellaert

Eindhoven, December 17, 2021


Contents

1 Week 1 1
1.1 Topic 1: Problem Description . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2 Week 2 4
2.1 Topic 2: Forecasting new patients . . . . . . . . . . . . . . . . . . . . . . . . 4

3 Week 3 7
3.1 Topic 3: Operating theatre planning . . . . . . . . . . . . . . . . . . . . . . 7

4 Week 4 10
4.1 Topic 4: Forecasting existing patients . . . . . . . . . . . . . . . . . . . . . 10

ii
1 Week 1
1.1 Topic 1: Problem Description
In every research, there are some time related costs and spending that time on research
instead of the company’s main operations should be well motivated. This is where the
problem description comes in. In this section, it should become clear what the exact prob-
lem is and why this problem is that relevant and worthwhile to persuade the start of a
research. Therefore, the problem description is a critical research element.

To the problem description of Jeroen den Hollander (24) some remarks can be given.
As discussed in class, Jeroen den Hollander’s problem description clarifies that he wants
to admit more patients in order to decrease the fixed costs while the bed utilization is
already high. However, the way he describes his problem creates some confusion and is
contradictory. In his thesis, he states that the hospital aims for improving their financial
status by admitting more patients to decrease the fixed costs which would logically be due
to low bed utilization. However, he tells us that the patient demand already surpasses the
bed capacity. Therefore, admitting more patients is impossible His explanation could be
improved by describing the gap between the actual problem and the desired goal. More-
over, he does not explain why this problem is a problem and why it is important to be
solved. At the end of his problem description, he mentions three problems in the form
of questions, which is an unusual way of formulating a problem. The content of these
questions are not even mentioned in the paragraphs before. So for Jeroen den Hollander,
writing a more structured problem description helps his audience to understand the real
problem better. Although, the length of his problem description is quite short, still some
repetition is present. Therefore, some parts can be written more concise.

An example of a thesis that provides a more clear problem description is that of Car-
lijn Goedhart (29). In the first paragraph, she is very concise in her explanation. She
clearly describes that the main problem is the dissatisfaction of the patient about the
time period of the surgery date announcement. Then, she elaborates this problem with
three sub-problems which are explained in more detail. Compared to Jeroen den Hollan-
der, the clear structure of Carlijn Goedhart’s problem description aids to understand the
problem. Except from the last paragraph, here the third sub-problem is explained using
too many causes. Therefore this paragraph is less concise compared to the others. One
class member mentioned that due to all the extra causations of each sub-problem, it is
not clear what the focus of her thesis is, which I agree upon. Although, motivating why
the problem is a problem and why it is crucial to be solved, are important elements of a
problem description, Carlijn Goedhart does not mention them.

Compared to the thesis of Carlijn Goedhart and Jeroen den Hollander, the way Gabrielle
van Nes (10) describes her problem is very concise and to the point. She explains very
clearly that there is a gab in the literature, as there is not much literature on how to deal
with limited number of available hospital beds, especially when searching for a practical
implementation such as certain flexibility methods. Therefore, this is the subject she is
going to focus on in her thesis. However, she does not relate the problem to the problem
of the hospital who she writes the thesis for. Furthermore, she has a clear structure in her
problem description by first explaining the literature gap, and then what she would like
know more about. Unfortunately, she does not motivate why it is a problem and why it
is important to be solved.

1
The thesis of Henriette Tas (20) considers this aspect of why the described problem is
important to be solved very well. She tells us that because of modifications in the internal
and external environment, certain departments wish to have more effective and efficient
operations with regards to money, resources and patients. Subsequently, she explains that
the focus is on the hospital beds. Except from the argument that in general there is no
efficient usage of resources, Henriette Tas does not include why she chooses for hospital
beds specifically. Furthermore, by using a problem statement, she communicates the main
problem: “Too many patients are refused, while the utilization of the resources is not very
high.”. The way she writes this problem statement is not very neath. It would be better
if she articulates it as follows: “At UMCN there is a high refusal rate of patients while
the utilization of the resources is relatively low.”. Also, the way the problem description
is structured, is not very supportive in recognizing what the real problem is. Only the
problem statement makes this clear. At the end, Henriette Tas sums up what problems
are related to the problem description which makes it less concise.

An example of a problem description that is structured, is that of Lizanne Pieters (26).


She starts off by telling us why the problem is necessary to be solved which is for negoti-
ation reasons with the health care insurance. According to Lizanne Pieters, the problem
is an actual problem because the capacity usage per patient group should be known for
the negotiations, while it is actually not available at the moment. She, then, concludes
her problem description by stating the problem as a question. This clarifies what the
problem is, but it is uncommon to express the problem in this manner. In addition, she
includes the differences in meaning of various terms which is unnecessary and it decreases
the conciseness.

When looking at the problem description of my own bachelor end project some com-
ments can be made as well. First of all, the problem is described very clearly and in a
concise manner. At the company under consideration, the lead time is longer than neces-
sary. As this lead time is the main competitive advantage of the company it is important
to be solved. Furthermore, the lead can be much shorter, so the problem is also an actual
problem. However, one should note that the structure of the problem description could
be improved.

In table 1, the overview of ratings of the theses on each element of the problem description
is shown. The thesis of Lizanne Pieters is the only thesis that both states why the prob-
lem is important to be solved (importance motivated) and motivates why the problem is
a problem (well motivated). Furthermore, every theses considers the audience they write
it for, except from Gabrielle van Nes who does not relate her problem to the concerning
hospital. Furthermore, Carlijn Goedhart and Gabrielle van Nes provide a clear description
of the problem that is also very concise and has a pleasant structure. All in all, for me
the problem description of Lizanne Pieters is the best. Although she is not very concise
and clear in the beginning about her problem, it is very important to have a grounded
motivation why the problem is a problem and why it is crucial to be solved. As Lizanne
Pieters is the only thesis who includes both elements, I prefer her problem description.

2
Jeroen Carlijn Gabrielle Henriette Lizanne BEP
den Hollander Goedhart van Nes Tas Pieters
(24) (29) (10) (20) (26)
Clear description – + ++ - 0 ++
Well motivated – – – - + +
Importance – – – + + +
motivated
Structure - ++ ++ - + -
Conciseness - + ++ - – +
Audience + + - + + +

Table 1: Overview of theses for topic 1

3
2 Week 2
2.1 Topic 2: Forecasting new patients
In a hospital, the demand for health care can be a critical component to investigate. In or-
der to optimize schedules of nurses and doctors or to determine the utilization of beds, this
knowing this demand can be useful. Unfortunately, this is not always certain in advance
as, hospitals do not only provide care in a way of pre-arranged appointments but also
provide the so called emergency care whereby patients need unplanned care immediately.
Furthermore, the duration that a bed is utilized or the duration of an operation are un-
known as well. For these reasons, a suitable forecasting technique of new patients is crucial.

One thesis that uses a satisfactory forecasting technique is that of Marle Muselaers (30).
She predicts the number of realized nursing hours by using a multi linear regression model.
Her model is only built on one year of data due to some well-grounded reasons. The model
consists of various independent variables to predict the dependent variable. The indepen-
dent variables are, amongst others, the number of patients present during the shift and
the percentage of patients with an age equal to or greater than 70. Unfortunately, one
should note that Marle Muselaers does not give any motivation why she chooses these
independent variables. Moreover, there can be observed that her model considers the hol-
iday planning indirectly as well, since the predicted variable in this thesis is the number
of realized nursing hours. This is satisfactory. One member of the class also noted that
she accounts for the movement of patient to other wards as well. This is important for
the forecasting the amount of nursing hours for each ward. Furthermore, it is pleasant
to note that she uses a clustering method. She clusters some period variables that have
an equal significant effect on the number of patients. By using a clustering method, a
model becomes smaller and stronger at the same time and the complexity of the model,
thus, decreases. Finally, her clustering method results into four clusters which have each
another bed occupancy.

The same period variables that Marle Muselaers uses, are used by Tim Kragten (25)
as well. However, his aim is to determine whether the bed occupancy is influenced by
the seasonal and trend effects using a multi linear regression model. He adds a good sub-
stantiated argument for using this type of model namely that it tackles the problem of
holidays which are different each year. Unfortunately, in the model he does not use any
clustering technique while using the same period variables as Marle Muselaers. Therefore,
his model is unnecessary complex as some of the periods have similar effects on the number
of patients. For a stronger model, the periods should have been clustered. In his thesis, he
forecasts the maximum number of beds during the day by using the number of OR sessions
and by creating dummy variables that are related to seasonal factors, holidays, off-days,
trend in years and the day of the week. When looking more carefully to his approach, Tim
Kragten does not measure the real demand for beds. As an example, during holidays the
maximum bed occupation during a day might seem lower. This does not mean that there
is less demand for health care during the summer. However, it could mean that there
are less available resources because more doctors are on holidays during the summer. As
the aim of his thesis is reducing imbalances between demand and supply of bed capacity,
it is not good at all to measure the number of accepted patients in stead of the actual
demand. In all, this means that his data is influenced by the elective patients. Implicating
that, during a period, whereby there is a lower maximum bed occupancy, there are also
less scheduled patients. Moreover, it is good to notice that the regression model of Tim

4
Kragten is based on five years of data, which increases the model’s accuracy. Furthermore,
when comparing the thesis of Marle Muselaers and Tim Kragten, forecasting the amount
of nursing hours instead of the bed occupancy differs in measuring of the amount of care
that is necessary. As the class discussion made clear, not every patient needs the same
amount of care. When a patient enters a hospital, the bed care is often low, however,
after an operation the bed care is often much higher. By predicting the realized amount
of nursing hours this element is considered.

Another thesis that does not measure the real demand is that of Joost Menting (27).
His model contains a multi linear regression analysis to forecast the number of patients
per weak. He uses a continuous variable for the year, a dummy variable for if a quartile is
quartile three and a dummy variable if a week had four or five working days as indepen-
dent variables. By using the four or five working days in a week to explain the differences
between weeks, Joost Menting does not take into account holidays. As Joost Menting’s
model is already small, no clustering is necessary. To build his model, he uses four years
of historical data, therefore, a trend over the years can be shown.

When comparing the models of Tim Kragten, Marle Muselaers and Joost Menting to
the models of Windi Winasti (6) and Bregje van der Staak (1), Windi and Bregje use a
different forecasting technique namely Winters’ forecasting model. The reason for this is
the presence of seasonal and trend pattern found in the historical data. Using a forecasting
model like Winters, limits a model in terms of holidays as these are different each year.
The model of Bregje van der Staak is based on three years of historical data. Therefore
she is able to visualize a trend over the years. Windi Winasti’s model is only based on
one year of data which implies that she can not investigate this trend. Just like Marle
Muselaers, Windi Winasti aims to forecast the required number of nursing hours. How-
ever, Windi Winasti uses her forecasting model to determine the long term estimation
of the monthly patient admission first, which she then translates to the required nursing
hours. For predicting the required nursing hours it is insufficient to only take into account
the seasonal and trend effects as there are also other variables that predict the required
nursing hours, for example the age of patients.

In table 2, the overview of the ratings on several criteria is shown. All theses aim at
forecasting real patient demand, but instead predict the amount of incoming patients or
nursing hours, which is not the same phenomenon. Furthermore, the models of Bregje
van der Staak, Windi Winasti and Joost Menting look only at different seasonal affects
and the high level trend over the years when forecasting the number of incoming patients.
However, as holidays differ each year, their models do not consider them. Furthermore,
by using the Winters’ forecasting model, it is not even possible to take other variables into
consideration. Although, Marle Muselaers does not have a well-grounded argument for
the use of her independent variables, I prefer her thesis the most, because she uses a multi
linear model with different independent variables. In addition, her dependent variable is
the realized nursing hours which contains more information than predicting the utilization
of bed as the amount of care a patient needs is taken into consideration as well.

5
Tim Marle Bregje Windi Joost
Kragten Muselaers van der Winasti Menting
(25) (30) Staak (1) (6) (27)
Years of historical data ++ 0 ++ - ++
Parameter optimization + ++ – - -
Motivation for + - + + +
parameter optimization
Appropriate model type + + - - +
Clustering method – ++ 0 0 0

Table 2: Overview of theses for topic 2

6
3 Week 3
3.1 Topic 3: Operating theatre planning
In a hospital, various resources have to be scheduled. The master surgery schedule (MSS),
for example, assigns departments to operation rooms (OR) and days. Also, multiple de-
cisions regarding the scheduling method have to be made at various organizational levels.
Examples of such decisions are, determining when a surgery is cancelled,or determining
how to take into account the non-elective patients in the planning. All of this belong the
operating theatre (OT) planning.

Carlijn Goedhart (29) analyses the OT at Radboudumc. She investigates whether it


is possible to schedule the surgery date at the moment there is decided that surgery is
necessary. For the hospital planner, it all starts by deciding upon the patient groups. In
the old method, patients who have similar time until the surgery are grouped together,
which is, thus, based on their medical similarity. This decision is tactical or strategical.
During the class discussion, it became clear that defining the strategic level is not always
easy. As there are only 3 levels of decision making, it is hard to specify such level when
the decision is somewhere in between. Next, a tactical decision is made, namely determine
the specialist capacity. Then, there needs to be decided whether to use a block or an open
strategy for planning surgeries. In an open strategy, the planning is based on the request
of surgeries by doctors. While in a block strategy, certain blocks are specified, which are
reserved for surgeries at certain departments. In the thesis of Carlijn Goedhart a block
strategy is used. One class member noted that by using this strategy, some flexibility is
lost as there are already defined operating blocks for the departments. Furthermore, by
using this strategy the preference of a doctor for the type of surgeries on a particular day
is not considered and their holidays have to be known earlier. However, the advantages of
a block strategy are that it is very convenient for a specialist to know on what day they
have to operate. After this decision, the slots are created. Meaning that for each day, OR
time slots are made. This is a tactical decision. Next, cases are assigned to resources.
This can be interpreted as doctors that can be assigned to a specialism, but also as the
allocation of a patient to an operation date. Finally, the cases are assigned to slots which
is operational. Carlijn Goedhart uses a MSS to assign departments to surgery days and
OR rooms. The MSS is based on the expected arrival of certain patients. For a MSS, it
is important to consider that every urgent patient is served and that a certain percentage
of each patient grouped is helped as well. There are also some constraints in a MSS, such
as the capacity of ORs, the nursing hours and the amount of beds. One can observe that
the nursing hours are a soft constraint as they can work overtime, however, for the beds
it is impossible to exceed the capacity which is thus a hard constraint.

In order for Carlijn Goedhart to analyze whether it is possible to choose the surgery
date of a patient at the moment that surgery is decided, the old OT planning, just ex-
plained, is redesigned. In the redesigned method, the first step is to find out the OR and
surgeon availability. In this step, she assumes that the MSS is already there, and she
checks whether it is still feasible when adding the surgeon availability. Then, the operat-
ing groups are defined which are based upon the amount of resources they use instead of
medical similarity. As these groups have similar operating duration, it makes it easier to
plan operations for each group. In the redesigned method, the first two steps of the old
method are swapped. For me, it would make more sense if defining the operating groups
is the first step as this decision is strategic. In the next step, there is checked whether the

7
capacity meets the demand. Especially when the actual incoming patients vary from the
predicted incoming patients or when capacity is temporarily lower, the MSS is modified.
This is a tactical decision. In the old method, these decisions are made based upon the
observed operational problems. Therefore, Carlijn Goedhart makes a good adjustment
to this method, because she prevents observing the problems too late. Then, the time
blocks are reserved which is similar to the old method. Next, the patients are admitted,
meaning that cases are assigned to sessions. A patient is allowed for admission when the
expected end time of the surgery is before 16:00, which is the same as before. Then, there
is dealt with operation cancellation. As it is possible that there is a difference between the
planned and the actual duration of an operation, it is crucial that there are rules for deal-
ing with operation cancellation. In the case of Carlijn Goedhart, only operations whereby
the expected end time is after 16:00 are cancelled. In the final step, there is dealt with
discrepancies between demand and supply. In the old method, this step is not necessary
to apply because patients can be switched up until one week in advance. One should note
that for every step that is changed, Carlijn Goedhart clearly adds well-grounded argu-
ments for it.

In order to determine, what type of block scheduling strategy is the best, four different
types are analysed. These are reserving no blocks, only reserving blocks for the operations
group, only reserving blocks for the surgeons and reserving blocks for both operations
group and surgeons. When there are no blocks reserved, a lower OR occupancy and more
patient cancellations are expected. However, the advantage is that it is more flexible and
has less overtime. When the operations groups are reserved in blocks, the risk of overtime
is limited as the operations groups are determined by their resource usage. By assigning
one specific surgeon to the same type of operations on a day, the operating duration is
expected to reduce as the surgeries contain repetition. When there is no reserved block for
the surgeons, a scheduled delay should be added because this can increase the probability
that the second surgery is started on time. The actual results of the simulation show that
the results for the four strategies are quite similar. However, when both operation groups
and surgeons are reserved in a block, the most improvement is shown with respect to, for
instance, acceptability and occupancy. Carlijn Goedhart considers the preference of the
doctors as well, when choosing the final strategy. For surgeons it is preferred to reserve
time blocks for them. Therefore, she chooses to reserve blocks for both operations groups
and surgeons which I agree upon.

Then, five combinations of admission and cancellation rules are analysed. The admis-
sion rule means that it is not allowed to plan operations that are expected to finish after
some pre-specified time, which are 8 or 8.5 hours in her analysis. When during the day,
there is expected that the finish time of a surgery is later than the time in the cancellation
rule, the surgery is cancelled. A variation between the admission and cancellation rule
allows for a softer constraint regarding the expected finish time of a surgery as some over-
time is allowed. Therefore, less last minute cancellation of surgeries are expected. The
results show that for all the five combinations, the waiting list strongly decreases. I would
expect that adding half an hour extra to the admission rule as the cancellation rule, would
reduce the average waiting time significantly. However, according to the results this is not
always the case. When deciding which combination of rules is the best, different trade-offs
need to be made. When adding some extra time to the cancellation rule, the nurses and
doctors have to work more overtime, whereby some resistance could occur. Therefore, the
nine hour cancellation rule is definitely not recommended. Furthermore, when there is a

8
8 hour admission rule and 8 hours cancellation rule, compared to the other combinations
whereby the cancellation rule is 8 or 8.5 hours, There can be seen that the percentage of
the last operations that are cancelled, are already relatively low. When keeping in mind
the possible resistance, I would recommend to have a 8 hour admission rule and cancella-
tion rule.

Ezgi Cayiroglu (11) also researches the OT planning as his aim is to schedule the op-
erations for each patient category. In his objective function he minimizes the overuse of
resources and the deviations of utilization levels from the utilization target levels. There-
fore, he limits the over and under usages of capacity. Also, a penalty is added for the
overuse of a resource compared to the capacity. In his constraints, there is a difference
between the overuse and the overutilization of a resource. When an OT is overused it
means that the OT’s capacity is exceeded. As this is a hard constraint, a high penalty is
assigned to it. When the OT is overutilized it means that it exceeds its target utilization,
which is a soft constraint. Furthermore, one class member noted that there is an error in
constraint 4, as the sign should be turned.

Another element of the OT planning is the optimal capacity allocation and planning
policy for non-elective patients, which Tim de Keijzer (12) aims to give insights on. In
order to investigate this, he uses five different scenarios regarding the allocation of capac-
ity over the OR day and three different planning policies which are both represented in
useful figures. In my opinion, scenario 2 where the capacity is allocated over the end of
the OR day with a peak at the end of the OR-day is the best scenario because the queue
length depends on the number of arriving non-elective patients. As there is expected that
this is the highest at the end of the OR-day, the capacity peak is added here. Further-
more, the flexible policy has my preference, as for non-elective patients the duration of an
operation can vary much. Using a flexible policy whereby multiple ORs can be used, the
waiting time of non-elective patients can be reduced. Furthermore, the utilization of ORs
is expected to be higher, because there is not one OR specifically assigned for non-elective
patients. However, one should note that more surgeries of elective patients will probably
be cancelled when using the flexible policy. The results of Tim de Keijzer show that the
scenario where the capacity is allocated over the end of the OR-day and evening with a
peak at the and of the OR-day and the scenario where it is allocated to end of the OR-day
and begin of the evening, perform the best regarding the average number of patients in
the queue. Especially when these two scenarios are combined with the mixed policy.

All in all, it is important to have good OT planning as it impacts many different as-
pects, such as the non-elective patients in the thesis of Tim de Keijzer, the overtime
worked by doctors in the thesis of Carlijn Goedhart, or the resource usage in the one
of Ezgi Cayiroglu. Finally, there should be noted that all these theses are based upon
the historical data. When the patient arrivals are completely different from the historical
data, it is crucial for the model to have clear rules how to adjust to it. As for example
Carlijn Goedhart did by adjusting the patient groups every year.

9
4 Week 4
4.1 Topic 4: Forecasting existing patients
Surgeries and consultations are together in an ongoing cycle. In a hospital, the amount of
surgeries in a week depends on the number of consultations in the weeks before. In turn,
the number of consults is determined by the amount of executed surgeries in the weeks
before. Therefore, when the number of patients that have a consult this week is known,
there can be forecasted how many patients need a surgery in the upcoming weeks.

An example of a thesis that forecasts the number of existing patients is that of Mirjam
Peters (32). In the department that is under investigation, patients stay relatively long.
Therefore, it is crucial to predict how many patient will be in the different units when
there is known how many patients there are at each unit at the moment. In her model,
she assumes that the length of stay (LOS) is geometrically distributed. One member of
the class noted that this distributions contains the memoryless property. Meaning that,
the duration of a patient in a certain unit is independent of the already spent duration
in that unit. While in reality, the probability that a patient is staying for some time at a
unit decreases when the patient has already spent some time there. Therefore, it would be
appreciated if Mirjam Peters would be critical about her assumption. In a Markov-chain,
she shows the amount of transitions between units and the number of patients flowing
in and out the system. From these numbers, a transition matrix is made, which gives
a good representation of the probability of going from one unit to another. In order to
obtain this matrix, first the likelihood of staying in the same unit is calculated using the
expected LOS. Then, the chance for transitioning from one unit to another is calculated
by multiplying the probability that a patient does not stay at its current unit with the
probability that the patient transitions from that unit to another specific unit. Finally,
she calculates the expectation and variance of the number of patients per unit per shift
using the formulas of the binomial distribution. For calculating this, the probability of
transitioning from one unit to another during a certain shift, decreases exponentially with
the number of shifts. Therefore, she takes into account that the likelihood of transitioning
of a patient that has already stayed some time at a unit, decreases. As this resembles
reality, this method is very pleasant.

In the thesis of Bregje van der Staak (1) an automatic resource model is built using a
Discrete-Time Markov Chain (CTMC). In her model a delay is added to the CTMC as
patients move from one state to the other with some delay. This delay between states is
taken into account when calculating the number of patients waiting in the queue at a cer-
tain state. This is the number of patients that arrived before plus the number of patients
that transitioned from an earlier state, while considering the delay, that are transferred to
this specific state. For every period the waiting list is calculated again using the waiting
list of the period before. Furthermore, compared to Mirjam Peters, Bregje van der Staak
does not take into account that the probability of staying any longer at a certain state
is dependent on the time that is already spent at that state. Therefore, as time passes,
still with the same probability patients move from one state to another. This is not very
satisfactory as it does not represent reality.

The same can be said about the thesis of Lona van Ruitenburg (8). In her model the
probability of having a surgery is not dependent on the number of surgeries a patient has
already had. When looking at the reality, the probability of having a second surgery is

10
much smaller. Therefore, one should be critical about her model. Furthermore, during
the class discussion various differences where noted between the flow diagram of Mirjam
Peters and the one of Lona van Ruitenburg. The representation of Lona van Ruitenburg
only shows percentages, therefore no absolute numbers of the amount of patients flowing
in are given. Also, one can notice that in Mirjam Peters’ figure, the lines are thicker when
more patients flow between these departments. This is not present in the figure of Lona
van Ruitenburg as well. When investigating the model definition of Lona van Ruitenburg
in more detail, in the objective function the number of waiting patients over the year is
minimized. The waiting patients consists of patients waiting for a consult and patients
waiting for an operation. In the objective function, these two variables are implemented
using a weighting factor which represents the relative importance. The model decreases
the queues for operations and consults by adjusting the number of sessions in the operating
rooms (OR) and outpatient departments (OD). Moreover, when defining her model, differ-
ent constraints about consults and operations are explained. The transition probabilities
reappear in these constraints. For example, the number of patients waiting for an opera-
tion with a certain urgency level in a certain period contains out of the expected number
of patients that need an operation after being at the OD and the number of patients that
is expected to have an extra operation. For the first element the probability that a patient
flows from the out patient department to the OR is needed and for the second element
the probability that a patient stays in the OR is necessary. Furthermore, the number
of patients waiting for a consult in the OD are also determined by the probabilities of
patients flowing to the OD.

All in all, when investigating the transitions of existing patients, the model of Mirjam
Peters is preferred compared to the models of Lona van Ruitenburg and Bregje van der
Staak. The reason for this is that in her model, the number of patients at certain unit
during a certain shift are not only dependent of the number of patients already at that
unit and the probability of transitioning from one unit to another, but also on the number
of shifts that have passed. This is an important element when comparing the model to
the real life situation.

11

You might also like