project management in healthcare: a case for evaluating patient flow of an emergency department with...
TRANSCRIPT
YILDIZ TECHNICAL UNIVERSITY MECHANICAL FACULTY
INDUSTRIAL ENGINEERING DEPARTMENT
PROJECT MANAGEMENT IN HEALTHCARE: A CASE FOR
EVALUATING PATIENT FLOW OF AN EMERGENCY DEPARTMENT
WITH FUZZY CPM AND FUZZY PERT
GİZEM GÜNEŞ 11061007
ADVISOR
Assoc. Prof. Ali Fuat GÜNERİ
ISTANBUL, 2015
1
PROJECT MANAGEMENT IN HEALTHCARE: A CASE FOR
EVALUATING PATIENT FLOW OF AN EMERGENCY
DEPARTMENT WITH FUZZY CPM AND FUZZY PERT
GİZEM GÜNEŞ
ADVISOR
Assoc. Prof. Ali Fuat GÜNERİ
Abstract
Nowadays projects are more complex than being in the past and complexibility increases
uncertainty. Fuzzy numbers are more effective for high uncertainty situations. In that study,
fuzzy numbers will be explained and then, how you can use fuzzy numbers in a project will
be explained with Fuzzy CPM (Critical Path Method) and Fuzzy PERT (Program Evaluation and
Review Technique). These tools provides to analyze flows and forecast project completing
time. In healthcare sector, FCPM and FPERT provide to standardize care, improve quality and
reduce patient’s staying time despite uncertainty.
Yıldız Technical University, Industrial Engineering Department, Istanbul, 2015
2
CONTENTS
1. Introduction …………………………………………………………………………………………….……………………… 3
2. Fuzzy Numbers ……………………………………………………………………………………………………………….. 5
2.1. Arithmethic Operations Between Fuzzy Numbers ……………………………………………..………………………… 6
3. Fuzzy Critical Path Method ……..………………………………………………………………………………………. 6
3.1. Notation ……………………………………………………………………………………………………………………………………… 6
3.2. Steps of Fuzzy CPM ……………………………………………………………………………………………………………………… 7
4. A Case of Critical Path Method ………..……..………………………………………………………………………. 8
5. Fuzzy Program Evaluation and Review Technique ( FPERT ) ………..……..…………………………. 16
6. Conclusion ..…………………………………………………………………………………………………………………… 18
7. References ……………………………………………………………………………………………………………………. 19
3
1. INTRODUCTION
In the past, uncertainity was less because projects were less complex. But now,
globalization improves day by day so, when anything is changed companies want to know
how it can affect result. Nowadays, information is power, so if the companies want to be
successful they have to have their past data to forecast their future. If the companies
analyze their past data they will grab the information; but, if they don’t they will look at the
flying information. If we specialize the issue for managing a project in healthcare sector
results are more important than other projects’. Because time is important but, in hospitals
time is the most important.
Healthcare Project Management Literature : L.C. Lee et al. [6] used PERT/CPM tools on
pulmonary lobectomy patients’ processes. provides to standardize care, reduce the lenghth
of stay and apply quality control processes to improve patients care with PERT/CPM. V.R.
Girija and M.S. Bhat [7] collected data with sample of 100 patients randomly selected from a
population of 460 and they researched to find critical path, expected completion time,
variance of the path and different slack timings associated with the emergency care
provided in the emergency department. U.K.M. Teichgräber et al. [8] used critical pathway
method to design and improve work flow in computed tomography.
While a Project is planning, you can use different kinds of data, these are : deterministic or
probabilistic, in this study fuzzy concept will be used for project management. Firstly, fuzzy
numbers will be explained and then, how you can use fuzzy numbers in a project while you
are planning, managing and controlling a project will be explained. While a project is
planned and managed, some tools are used such as CPM (Critical Path Method) and PERT
(Program Evaluation and Review Technique). These tools are effective and simple methods
to analyze flows and forecast ending date of a project. But, before these tools are used, right
collected database and right method of analyzing data has to be prepared.
4
Fuzzy Project Management Literature : Zadeh [9] researched relationship between
possibility and fuzzy numbers. G.S. Liang and T.C. Han [1] used an algorithm that is related to
decision maker’s attitude (pessimistic/optimistic); Shankar et al. [2] used a defuzzification
algorithm; N.R. Shankar et al. [3] proposed a metric distance method for trapezoidal fuzzy
numbers and found float time for each activity to find the critical path in the fuzzy project
network. S.P. Chen and Y.J. Hsueh [10] proposed a method that based on linear
programming formulation and fuzzy number ranking method. M.F. El-Santaway and M. Abd-
Allah [4] used a linear programming approach which incorporates the concept of α-cuts into
two (primal and dual) models to identfy the critical path. V. Sireesha et al [5] proposed a
graphical method of solving fuzzy interval CPM problem. M. Sharafi et al. [12] used a linear
programming model for triangular fuzzy numbers to find earliest time, latest time and slack
times of a project. N.S. Pour [13] studied a method that doesn’t use any defuzzification
technique to find the final processing time.
In that study, fuzzy sets will be explained and then, how you can use fuzzy numbers in a
project will be explained with Fuzzy CPM (Critical Path Method) and Fuzzy PERT (Program
Evaluation and Review Technique) tools with a case. The case is about processes of patients
in emergency department. Sample size is 1500 patients. In the case, earliest and latest fuzzy
times and total slack time will be found with using FCPM and FPERT tools. In healthcare
sector, FCPM and FPERT provide to standardize care, improve quality and reduce patient’s
staying time. In addition, slack times between related activities can be found with
FCPM/FPERT tools and they provide to analyze space time and improve productivity in the
project.
5
2. FUZZY NUMBERS
A fuzzy number A in R (Real Line) is a trapezoidal fuzzy number is its membership function
FA (x) : R → [0, 1] is
with - ∞ ≤ a ≤ b ≤ c ≤ d ≤ ∞, the trapezoidal fuzzy number A can be represented by (a, b, c,
d) (Liang and Han, 2004). Using this function, it is possible to assign a membership degree to
each of the element in the universe of discourse X. It is important to note the fact that
membership grades are not probabilities. (Shankar, Sireesha, Rao and Vani, 2010) It is
shown in Figure 1.
A fuzzy number represented by (a, b, c,
d). a means minimum number of data set;
b and c means range of average of data
set; d means maximum number of data
set. For example, we think a fuzzy number
is (12, 16, 18, 20). It can be minimum at 12
and maximum at 20. It can be
approximately between 16 and 18. It is
shown in Figure 2.
Figure 1. Crisp Set and Fuzzy Set (Shankar, Sireesha, Rao and Vani, 2010)
Figure 2. Trapezoidal Fuzzy Number (12, 16, 18, 20)
a = 12 b = 16 c = 18 d = 20
X
1
0
µA (X)
6
If b equals to c that means it can be
approximately number of b = c. For
example (2, 4, 4, 6) means it can be
minimum at 2 and maximum at 6; and it can
be approximately 4. It is shown in Figure 3.
2.1. Arithmetic Operations Between Fuzzy Numbers
For A1 = (a1, b1, c1, d1) and A2 = (a2, b2, c2, d2),
Addition : A1 ⃝ A2
A1 ⃝ A2 = (a1, b1, c1, d1) ⃝ (a2, b2, c2, d2) = (a1 + a2, b1 + b2, c1 + c2, d1 + d2)
Subtraction : A1 ⃝ A2
A1 ⃝ A2 = (a1, b1, c1, d1) ⃝ (a2, b2, c2, d2) = (a1 - d2, b1 - c2, c1 - b2, d1 - a2)
Examples
Let A1 and A2 be fuzzy numbers which are (8, 9, 10, 11) and (3, 4, 5, 6).
A1 ⃝ A2 = (8, 9, 10, 11) ⃝ (3, 4, 5, 6) = (8 + 3, 9 + 4, 10 + 5, 11 + 6) = (11, 13, 15, 17)
A1 ⃝ A2 = (8, 9, 10, 11) ⃝ (3, 4, 5, 6) = (8 – 6, 9 – 5, 10 – 4, 11 – 3) = (2, 4, 6, 8)
3. FUZZY CRITICAL PATH METHOD ( FCPM )
3.1. Notation
N : All nodes in a project network
Aij : The activity between two nodes (from i to j)
FATij : Fuzzy activity time of Aij
EFT : Earliest fuzzy time
LFT : Latest fuzzy time
SFTij : Total slack fuzzy time of Aij
FCPM (Pn) : Fuzzy completion time of n th path
t : Number of activities in a project network
X
a = 2 b = c = 4 d = 6
1
0
µA (X)
Figure 3. Triangular Fuzzy Number (2, 4, 4, 6)
+ +
+
-
- -
+ +
- -
7
3.2. Steps of Fuzzy CPM
Step 1. Accept EFT1 equals to (0, 0, 0, 0).
Step 2. Calculate β that means risk factor for each Aij with these formulation :
β = ∑ ∑ 𝐛𝐢𝐣−𝐚𝐢𝐣
(𝐛𝐢𝐣−𝐚𝐢𝐣)+(𝐝𝐢𝐣−𝐜𝐢𝐣) / t
Step 3. Calculate earliest fuzzy time (EFT) for each node.
EFTj = EFTi ⃝ FATij
Step 4. Compare EFTj s where intersection nodes and accept maximum number for EFTj for
each node.
EFTj = max { EFTi ⃝ FATij }
Step 5. Calculate latest fuzzy time (LFT) for each node.
LFTj = EFTk ⃝ FATjk
Step 6. Compare LFTj s where intersection nodes and accept minimum number as LFTj for
each node.
LFTj = min { EFTk ⃝ FATjk }
+
+
EFTj = max {(ax, bx, cx, dx), (ay, by, cy, dy)}
Step 4.1. Find x1 and x2 values with these equations :
x1 = min { ax, bx, cx, dx, ay, by, cy, dy} , x2 = max { ax, bx, cx, dx, ay, by, cy, dy}
Step 4.2. Calculate R ((ax, bx, cx, dx)) and R((ay, by, cy, dy)) values with these equation :
R ((ai, bi, ci, di)) = β [(di - x1)/( x2 - x1 - ci + di)] + (1 – β) [1 – (x2 - ai) / ( x2 - x1 + bi - ai)]
Step 4.3. Compare results of R ((ai, bi, ci, di)) and accept as EFTj which one is greater.
-
-
( ) i j
Note : β ˂ 0,5 , it’s fewer risky situation
If β = 0,5 , it’s neutral situation
β ˃ 0,5 , it’s risky situation
8
Step 7. Calculate Total Slack Fuzzy Time (SFT) for each activity.
SFTij = LFTj ⃝ ( EFTi ⃝ FATij)
Step 8. Find all possible path and calculate Fuzzy Completion Time of Path (FCPM) for each
one. FCPM can be calculated with sum of activities between possible path nodes.
FCPM (Pn) = ∑ SFTij
Step 9. Compare FCPM s and accept minimum number as FCPM just like Step 6 calculation.
FCPM (Pn) = min { FCPM (Pi) | i = 1, 2, 3 … , n }
Step 10. Which FCPM of path is fewer, accept the path as critical path.
4. A CASE OF FUZZY CRITICAL PATH METHOD
This example is about patient’s flow in emergency department. The patient arrives the
emergency department by ambulance or patient’s private vehicle and patient’s is recorded
before being examined by doctor. If patient arrives by ambulance the patient passes
recording operation, so the patient goes to near of doctor directly. After doctor examines
the patient who transfers the patient to available part of emergency department depends
on patient’s situation. A sample is taken from the patient and performing lab tests. In same
time, the patient is got under control by nurse and the doctor. The patient can exit with 3
different ways depends on result of patient’s lab tests and doctor recommendation which
are : 1. the patient can leave hospital, 2. the patient can be transfered to another hospital, 3.
the patient can be shifted to appropriate department depends on patient’s situation. In the
case, sample size is 1500 patients.
LFTj = min {(ax, bx, cx, dx), (ay, by, cy, dy)}
Step 6.1. Find x1 and x2 values with these equations :
x1 = min { ax, bx, cx, dx, ay, by, cy, dy} , x2 = max { ax, bx, cx, dx, ay, by, cy, dy}
Step 6.2. Calculate R ((ax, bx, cx, dx)) and R((ay, by, cy, dy)) values with these equation :
R ((ai, bi, ci, di)) = β [(di - x1)/( x2 - x1 - ci + di) + (1 – β) (1 – (x2 - ai) / ( x2 - x1 + bi - ai)
Step 6.3. Compare results of R ((ai, bi, ci, di)) and accept as EFTj which one is fewer.
- +
9
Patient’s activities in emergency department is shown in Table 1, project network is shown
in Figure 4 and activities’ times are shown in Table 2.
Symbol Activity
A Shifting of the patient from ambulance to emergency department
B Shifting of the patient from private vehicle to emergency department
C Recording of the patient
D Examining of the patient by doctor
E Sending of the patient to available part of the emergency department
F Checking of the patient by doctor and nurse
G Performing lab tests
H Shifting of the patient to appropriate department of the hospital depends on patient's situation
I Registration
J Shifting of the patient to ambulance to transfer another hospital
K Billing
Table 1. Activities of Patient in Emergency Department
Table 2. Activities’ Times of Patient in Emergency Department
1
2
A 3 4
B C
D
6
5E
F
G 7
H
8J
9K
10I
Figure 4. Project Network of Patient’s Activities in Emergency Department
Activity Predecessor a b c d
A - 0,5 1,2 1,5 4
B - 1 1,8 2,1 4
C B 0,7 1 1 2,5
D A, C 2,5 5 7 9
E D 0,4 0,5 0,8 1,4
F E 8 15 28 134
G E 1 10 10 125
H G, F 5 6 8 9
I G, F 5 8 8 14
J I 1 2,1 2,5 5
K I 5 6 6 8
10
Step 1. EFT1 = (0, 0, 0, 0)
Step 2. β = 0,2715 is found with that equation:
Step 3. Calculate EFT s for each node with that equation:
EFTj = EFTi ⃝ FATij
Step 4. Compare EFT s for each node and accept maximum one as EFT.
EFT2 = (0, 0, 0, 0) + (1, 1,8 , 2,1 , 4) = (1, 1,8 , 2,1 , 4)
EFT3 = max {(EFT2 ⃝ FAT23) , (EFT1 ⃝ FAT13)}
EFT3 = max {((1, 1,8 , 2,1 , 4) ⃝ (0,7 , 1, 1, 2,5)), ((0, 0, 0, 0) ⃝ (0,5 , 1,2 , 1,5 , 4))}
EFT3 = max {(1,7 , 2,8 , 3,1 , 6,5) , (0,5 , 1,2 , 1,5 , 4)}
R((1,7 , 2,8 , 3,1 , 6,5)) = 0,4093, R((0,5 , 1,2 , 1,5 , 4)) = 0,1879
EFT3 = (1,7 , 2,8 , 3,1 , 6,5)
EFT4 = EFT3 ⃝ FAT34
EFT4 = (1,7 , 2,8, 3,1 , 6,5) ⃝ (2,5 , 5, 7, 9)
EFT4 = (4,2 , 7,8 , 10,1 , 15,5)
EFT5 = (4,2 , 7,8 , 10,1 , 15,5) ⃝ (0,4 , 0,5 , 0,8 , 1,4)
EFT5 = (4,6 , 8,3 , 10,9 , 16,9)
EFT6 = EFT5 = (4,6 , 8,3 , 10,9 , 16,9)
EFT7 = max {(EFT6 ⃝ FAT67) , (EFT5 ⃝ FAT57)}
EFT7 = max {((4,6 , 8,3 , 10,9 , 16,9) ⃝ (8, 15, 28, 134)) , ((4,6 , 8,3 , 10,9 , 16,9) ⃝ (1,
10, 10, 125))}
EFT7 = max {(12,6 , 23,3 , 38,9 ,150,9) , (5,6 , 18,3 , 20,9 , 141,9)}
R((12,6 , 23,3 , 38,9 ,150,9)) = 1,9933, R((5,6 , 18,3 , 20,9 , 141,9)) = 1,0911
EFT7 = (12,6 , 23,3 , 38,9 , 150,9)
EFT8 = EFT7 ⃝ FAT78
EFT8 = (12,6 , 23,3 , 38,9 , 150,9) ⃝ (5, 8, 8, 14)
EFT8 = (17,6 , 31,3 , 46,9 , 164,9)
EFT9 = EFT8 = (17,6 , 31,3 , 46,9 , 164,9)
EFT10 = max {(EFT7 ⃝ FAT710) , (EFT8 ⃝ FAT810) , (EFT9 ⃝ FAT910)}
EFT10 = max {((12,6 , 23,3 , 38,9 , 150,9) ⃝ (5, 6, 8, 9)) , ((17,6 , 31,3 , 46,9 , 164,9) ⃝
(1, 2,1 , 2,5 , 5)) , ((17,6 , 31,3 , 46,9 , 164,9) ⃝ (5, 6, 6, 8))}
EFT10 = max {(17,6 , 29,3 , 46,9 , 159,9) , (18,6 , 33,4 , 49,4 , 169,9) , (22,6 , 37,3 , 52,9
, 172,9)}
R((17,6 , 29,3 , 46,9 , 159,9)) = 1,5490 , R((18,6 , 33,4 , 49,4 , 169,9)) = 1,5159 ,
R((22,6 , 37,3 , 52,9 , 172,9)) = 1,6666
EFT10 = (22,6 , 37,3 , 52,9 , 172,9)
+
+
+ +
+
+
+
+
+ +
+ +
+
+
+ + +
+ +
+
11
Step 5. Calculate LFT s for each node with that equation:
LFTj = EFTk ⃝ FATjk
Step 6. Compare LFT s for each node and accept minimum one as EFT.
LFT10 = (22,6 , 37,3 , 52,9 , 172,9)
LFT9 = LFT10 ⃝ FAT910
LFT9 = (22,6 , 37,3 , 52,9 , 172,9) ⃝ (5, 6, 6, 8)
LFT9 = (14,6 , 31,3 , 46,9 , 167,9)
LFT8 = min { (LFT9) , (LFT10 ⃝ FAT810 ) }
LFT8 = min { (14,6 , 31,3 , 46,9 , 167,9) , [(22,6 , 37,3 , 52,9 , 172,9) ⃝ (1, 2,1 , 2,5 , 5)]}
LFT8 = min { (14,6 , 31,3 , 46,9 , 167,9) , (17,6 , 34,8 , 50,8 , 171,9) }
R((14,6 , 31,3 , 46,9 , 167,9)) = 0,2195 , R((17,6 , 34,8 , 50,8 , 171,9)) = 0,2377
LFT8 = (14,6 , 31,3 , 46,9 , 167,9)
LFT7 = min { (LFT10 ⃝ FAT710) , (LFT8 ⃝ FAT78 ) }
LFT7 = min { [(22,6 , 37,3 , 52,9 , 172,9) ⃝ (5, 6, 8, 9)] , [(14,6 , 31,3 , 46,9 , 167,9) ⃝
(5, 8, 8, 14)] }
LFT7 = min { (13,6 , 29,3 , 46,9 ,167,9) , (0,6 , 23,3 ,38,9 , 162,9) }
R((13,6 , 29,3 , 46,9 ,167,9)) = 0,2718 , R((0,6 , 23,3, 38,9 , 162,9)) = 0,2383
LFT7 = (0,6 , 23,3 ,38,9 , 162,9)
LFT6 = LFT7 ⃝ FAT67
LFT6 = (0,6 , 23,3 ,38,9 , 162,9) ⃝ (8, 15, 28, 134)
LFT6 = (-133,4 , -4,7, 30,9 , 154,9)
LFT5 = min { (LFT6) , (LFT7 ⃝ FAT57 ) }
LFT5 = min { (-133,4 , -4,7, 30,9 , 154,9), [(0,6 , 23,3 ,38,9 , 162,9) ⃝ (1, 10, 10, 125)] }
LFT5 = min { (-133,4 , -4,7, 30,9 , 154,9) , (-124,4 , 13,3 , 28,9 , 161,9) }
R((-133,4 , -4,7, 30,9 , 154,9)) = 0,4048 , R((-124,4 , 13,3 , 28,9 , 161,9)) = 0,4340
LFT5 = (-133,4 , -4,7, 23,9 , 154,9)
LFT4 = LFT5 ⃝ FAT45
LFT4 = (-133,4 , -4,7, 23,9 , 154,9) ⃝ (0,4 , 0,5 , 0,8 , 1,4)
LFT4 = (-134,8 , -5,5 , 23,4 , 154,5)
LFT3 = LFT4 ⃝ FAT34
LFT3 = (-134,8 , -5,5 , 23,4 , 154,5) ⃝ (2,5 , 5, 7, 9)
LFT3 = (-143,8 , -12,5 , 18,4 , 152)
LFT2 = LFT3 ⃝ FAT23
LFT2 = (-143,8 , -12,5 , 18,4 , 152) ⃝ (0,7 , 1, 1, 2,5)
LFT2 = (-146,3 , -13,5, 17,4 , 151,3)
LFT1 = min { (LFT2 ⃝ FAT12) , (LFT3 ⃝ FAT13 ) }
LFT1 = min { [(-146,3 , -13,5, 17,4 , 151,3) ⃝ (1, 1,8 , 2,1 , 4)] , [(-143,8 , -12,5 , 18,4 ,
152) ⃝ (0,5 , 1,2 , 1,5 , 4)] }
LFT1 = min { (-150,3, -15,6, 15,6 , 150,3) , (-147,8 , -14, 17,2 , 151,5) }
-
-
-
-
-
- -
- -
-
-
-
-
-
-
-
-
-
-
-
-
-
-
12
R((-150,3, -15,6, 15,6 , 150,3)) = 0,4118 , R((-147,8 , -14, 17,2 , 151,5)) = 0,4159
LFT1 = (-150,3, -15,6, 15,6 , 150,3)
EFT and LFT values of each node are shown in Table 3.
Step 7. Calculate Total Slack Fuzzy Time (SFT) for each activity with that equation :
SFTij = LFTj ⃝ ( EFTi ⃝ FATij)
SFT13 = LFT3 ⃝ (EFT1 ⃝ FAT13)
SFT13 = (-143,8 , -12,5 , 18,4 , 152) ⃝ [(0, 0, 0, 0) ⃝ (0,5 , 1,2 , 1,5 , 4)]
SFT13 = (-147,8 , -14 , 17,2 , 151,5)
SFT12 = LFT2 ⃝ (EFT1 ⃝ FAT12)
SFT12 = (-146,3 , -13,5 , 17,4 , 151,3) ⃝ [(0, 0, 0, 0) ⃝ (1, 1,8 , 2,1 , 4)]
SFT12 = (-150,3 , -15,6 , 15,6 , 150,3)
SFT23 = LFT3 ⃝ (EFT2 ⃝ FAT23)
SFT23 = (-143,8 , -12,5 , 18,4 , 152) ⃝ [(1, 1,8 , 2,1 , 4) ⃝ (0,7 , 1, 1, 2,5)]
SFT23 = (-150,3 , -15,6 , 15,6 , 150,3)
SFT34 = LFT4 ⃝ (EFT3 ⃝ FAT34)
SFT34 = (-134,8 , -5,5 , 23,4 , 154,5) ⃝ [(1,7 , 2,8 , 3,1 , 6,5) ⃝ (2,5 , 5, 7, 9)]
SFT34 = (-150,3 , -15,6 , 15,6 , 150,3)
SFT45 = LFT5 ⃝ (EFT4 ⃝ FAT45)
SFT45 = (-133,4 , -4,7 , 23,9 , 154,9) ⃝ [(4,2 , 7,8 , 10,1 , 15,5) ⃝ (0,4 , 0,5 , 0,8 , 1,4)]
SFT45 = (-150,3 , -15,6 , 15,6 , 150,3)
SFT56 = LFT6 ⃝ EFT5
SFT56 = (-133,4 , -4,7 , 23,9 , 154,9 ) ⃝ (4,6 , 8,3 , 10,9 , 16,9)
a b c d a b c d
1 0 0 0 0 -150,3 -15,6 15,6 150,3
2 1 1,8 2,1 4 -146,3 -13,5 17,4 151,3
3 1,7 2,8 3,1 6,5 -143,8 -12,5 18,4 152
4 4,2 7,8 10,1 15,5 -134,8 -5,5 23,4 154,5
5 4,6 8,3 10,9 16,9 -133,4 -4,7 23,9 154,9
6 4,6 8,3 10,9 16,9 -133,4 -4,7 23,9 154,9
7 12,6 23,3 38,9 150,9 0,6 23,3 38,9 162,9
8 17,6 31,3 46,9 164,9 14,6 31,3 46,9 167,9
9 17,6 31,3 46,9 164,9 14,6 31,3 46,9 167,9
10 22,6 37,3 52,9 172,9 22,6 37,3 52,9 172,9
EFT LFT
- +
-
-
-
-
-
-
-
-
-
-
- +
+
+
+
+
+
+
+
+
+
-
Table 3. EFT s and LFT s values of each node
13
SFT56 = (-150,3 , -15,6 , 15,6 , 150,3)
SFT67 = LFT7 ⃝ (EFT6 ⃝ FAT67)
SFT67 = (0,6 , 23,3 , 38,9 , 162,9) ⃝ [(4,6 , 8,3 , 10,9 , 16,9) ⃝ (8, 15, 28, 134)]
SFT67 = (-150,3 , -15,6 , 15,6 , 150,3)
SFT57 = LFT7 ⃝ (EFT5 ⃝ FAT57)
SFT57 = (0,6 , 23,3 , 38,9 , 162,9) ⃝ [(4,6 , 8,3 , 10,9 , 16,9) ⃝ (1, 10, 10, 125)]
SFT57 = (-141,3 , 2,4 , 20,6 , 157,3)
SFT710 = LFT10 ⃝ (EFT7 ⃝ FAT710)
SFT710 = (22,6 , 37,3 , 52,9 , 172,9) ⃝ [(12,6 , 23,3 , 38,9 , 150,9) ⃝ (5, 6, 8, 9)]
SFT710 = (-137,3 ,-9,6 , 23,6 , 155,3)
SFT78 = LFT8 ⃝ (EFT7 ⃝ FAT78)
SFT78 = (14,6 , 31,3 , 46,9 , 167,9) ⃝ [(12,6 , 23,3 , 38,9 , 150,9) ⃝ (5, 8, 8, 14)]
SFT78 = (-150,3 , -15,6 , 15,6 , 150,3)
SFT89 = LFT9 ⃝ EFT8
SFT89 = (14,6 , 31,3 , 46,9 , 167,9) ⃝ (17,6 , 31,3 , 46,9 , 164,9 )
SFT89 = (-150,3 , -15,6 , 15,6 , 150,3)
SFT810 = LFT10 ⃝ (EFT8 ⃝ FAT810)
SFT810 = (22,6 , 37,3 , 52,9 , 172,9) ⃝ [(17,6 , 31,3 , 46,9 , 164,9 ) ⃝ (1 , 2,1 , 2,5 , 5)]
SFT810 = (-147,3 , -12,1 , 19,5 , 154,3)
SFT910 = LFT10 ⃝ (EFT9 ⃝ FAT910)
SFT910 = (22,6 , 37,3 , 52,9 , 172,9) ⃝ [(17,6 , 31,3 , 46,9 , 164,9 ) ⃝ (5, 6, 6, 8)]
SFT910 = (-153,3 , -15,6 , 15,6 , 153,3)
Step 8. Find all possible path and calculate Fuzzy Completion Time of Path (FCPM) for each
one. FCPM can be calculated with sum of activities between possible path nodes.
Let path P1 = (1 – 3 – 4 – 5 – 6 – 7 – 8 – 10) .
FCPM (P1) = SFT13 + SFT34 + SFT45 + SFT56 + SFT67 + SFT78 + SFT810
FCPM (P1) = (-1046,6 , -104,1 , 114,7 , 1057,3)
Let path P2 = (1 – 3 – 4 – 5 – 6 – 7 – 10) .
FCPM (P2) = SFT13 + SFT34 + SFT45 + SFT56 + SFT67 + SFT710
FCPM (P2) = (-886,3 , -86 , 103,2 , 908)
Let path P3 = (1 – 3 – 4 – 5 – 6 – 7 – 8 – 9 – 10) .
FCPM (P3) = SFT13 + SFT34 + SFT45 + SFT56 + SFT67 + SFT710
FCPM (P3) = (-1202,9 , -123,2 , 126,4 , 1206,6)
Let path P4 = (1 – 3 – 4 – 5 – 7 – 10) .
FCPM (P4) = SFT13 + SFT34 + SFT45 + SFT57 + SFT710
FCPM (P4) = (-727 , -52,4 , 92,6 , 764,7)
-
-
-
-
-
-
-
-
-
-
- +
+
+
+
+
+
+
+
+
+
+
+
-
-
-
14
Let path P5 = (1 – 3 – 4 – 5 – 7 – 8 – 10) .
FCPM (P5) = SFT13 + SFT34 + SFT45 + SFT57 + SFT78 + SFT810
FCPM (P5) = (-887,3 , -70,5 , 104,1 , 914)
Let path P6 = (1 – 3 – 4 – 5 – 7 – 8 – 9 – 10) .
FCPM (P6) = SFT13 + SFT34 + SFT45 + SFT57 + SFT78 + SFT89 +SFT910
FCPM (P6) = (-1043,6 , -89,6 , 115,8 , 1063,3)
Let path P7 = (1 – 2 – 3 – 4 – 5 – 6 – 7 – 8 –10) .
FCPM (P7) = SFT12 + SFT23 + SFT34 + SFT45 + SFT56 + SFT67 + SFT78 + SFT810
FCPM (P7) = (-1199,4 , -121,3 , 128,7 , 1206,4)
Let path P8 = (1 – 2 – 3 – 4 – 5 – 6 – 7 – 10) .
FCPM (P8) = SFT12 + SFT23 + SFT34 + SFT45 + SFT56 + SFT67 + SFT710
FCPM (P8) = (-1039,1 , -103,2 , 117,2 , 1057,1)
Let path P9 = (1 – 2 – 3 – 4 – 5 – 6 – 7 – 8 – 9 – 10) .
FCPM (P9) = SFT12 + SFT23 + SFT34 + SFT45 + SFT56 + SFT67 + SFT710
FCPM (P9) = (-1355,7 , -140,4 , 140,4 , 1355,7)
Let path P10 = (1 – 2 – 3 – 4 – 5 – 7 – 10) .
FCPM (P10) = SFT12 + SFT23 + SFT34 + SFT45 + SFT57 + SFT710
FCPM (P10) = (-879,8 , -69,6 , 106,6 , 913,8)
Let path P11 = (1 – 2 – 3 – 4 – 5 – 7 – 8 – 10) .
FCPM (P11) = SFT12 + SFT23 + SFT34 + SFT45 + SFT57 + SFT78 + SFT810
FCPM (P11) = (-1040,1 , -87,7 , 118,1 , 1063,1)
Let path P12 = (1 – 2 – 3 – 4 – 5 – 7 – 8 – 9 – 10) .
FCPM (P12) = SFT12 + SFT23 + SFT34 + SFT45 + SFT57 + SFT78 + SFT89 +SFT910
FCPM (P12) = (-1196,4 , -106,8 , 129,8 , 1212,4)
15
Step 9. Compare FCPM s and accept minimum number as FCPM.
FCPM and R (FCPM (Pi)) values of each possible path is shown in Table 4.
X1 = -1355,7 , X2 = 1355,7 , β = 0,2715
Step 10. Which FCPM of path is fewer, accept the path as critical path.
In a conclusion, R (FCPM (P9)) is minimum value of all possible paths’ FCPM s. So, the
critical path is 1 – 2 – 3 – 4 – 5 – 6 – 7 – 8 – 9 – 10 in that case. Length of treatment period is
approximately between 37,3 and 52,9 minutes i.e. (22,6 , 37,3 , 52,9 ,172,9).
a b c d
(1 – 3 – 4 – 5 – 6 – 7 – 8 – 10) -1046,6 -104,1 114,7 1057,3 0,4288
(1 – 3 – 4 – 5 – 6 – 7 – 10) -886,3 -86 103,2 908 0,4382
(1 – 3 – 4 – 5 – 6 – 7 – 8 – 9 – 10) -1202,9 -123,2 126,4 1206,6 0,4203
(1 – 3 – 4 – 5 – 7 – 10) -727 -52,4 92,6 764,7 0,4506
(1 – 3 – 4 – 5 – 7 – 8 – 10) -887,3 -70,5 104,1 914 0,4404
(1 – 3 – 4 – 5 – 7 – 8 – 9 –10) -1043,6 -89,6 115,8 1063,3 0,4311
(1 – 2 – 3 – 4 – 5 – 6 – 7 – 8 – 10) -1199,4 -121,3 128,7 1206,4 0,4209
(1 – 2 – 3 – 4 – 5 – 6 – 7 – 10) -1039,1 -103,2 117,2 1057,1 0,4296
(1 – 2 – 3 – 4 – 5 – 6 – 7 – 8 – 9 – 10) -1355,7 -140,4 140,4 1355,7 0,4130
(1 – 2 – 3 – 4 – 5 – 7 – 10) -879,8 -69,6 106,6 913,8 0,4412
(1 – 2 – 3 – 4 – 5 – 7 – 8 – 10) -1040,1 -87,7 118,1 1063,1 0,4317
(1 – 2 – 3 – 4 – 5 – 7 – 8 – 9 –10) -1196,4 -106,8 129,8 1212,4 0,4232
FCPM (Pi)R (FCPM(Pi))
Table 4. FCPM and R (FCPM (Pi)) values of each possible path
16
5. FUZZY PROGRAM EVALUATION AND REVIEW TECHNIQUE ( FPERT )
In that issue will be explained with previous case.
Step 1. Compute Mean for each activity with that equation :
m = ( a + b + c + d ) / 4
All means of activities are shown in Table 5.
Step 2. Compute Variance for each activity with that equation :
σ2 = [ 3 ( a2 + b2 + c2 + d2 ) – 2 ( ab + ac + ad + bc + bd + cd ) ] / 36
All variances of activities are shown in Table 6.
Activity a b c d Mean
A 0,5 1,2 1,5 4 1,800
B 1 1,8 2,1 4 2,225
C 0,7 1 1 2,5 1,300
D 2,5 5 7 9 5,875
E 0,4 0,5 0,8 1,4 0,775
F 8 15 28 134 46,250
G 1 10 10 125 36,500
H 5 6 8 9 7,000
I 5 8 8 14 8,750
J 1 2,1 2,5 5 2,650
K 5 6 6 8 6,250
Table 5. Mean Values of Each Activity
Table 6. Variance Values of Each Activity
Activity a b c d Mean Variance
A 0,5 1,2 1,5 4 1,800 0,7756
B 1 1,8 2,1 4 2,225 0,5386
C 0,7 1 1 2,5 1,300 0,2200
D 2,5 5 7 9 5,875 2,5764
E 0,4 0,5 0,8 1,4 0,775 0,0675
F 8 15 28 134 46,250 1163,6389
G 1 10 10 125 36,500 1166,3333
H 5 6 8 9 7,000 1,1111
I 5 8 8 14 8,750 4,7500
J 1 2,1 2,5 5 2,650 0,9522
K 5 6 6 8 6,250 0,5278
119,375 2341,4914
17
Note : ij is in Step 3 and 4 that means activities are on the critical path which has known.
And that case critical path includes all activities.
Step 3. Compute Standart Deviation of the Project with sum of variance of activities are on
critical path :
σ = ( ∑ σ2 ij )
1/2
In that case σ = ( 2341,4914 ) 1/2 = 48,389
Step 4. Compute Expected Project Completing Time with sum of mean of activities are on
critical path :
E ( Project ) = ∑ m ij
In that case m = 119,375
Step 5. To find probability of the project completed time is less than X value , Central Limit
Theorem will be used for it. Trapezoidal distribution is accepted to approximate normal
distribution. Normal distribution has two variables are mean and standart deviation and they
have been computed previous steps.
Z = ( X - µ ) / σ
Step 6. After Z value is computed, find probability with z table.
In that case, some probability calculation results are shown in Table 7.
< 80 < 100 < 120 < 140 < 160 < 180 < 200 < 220 < 240 < 280
Z value -0,814 -0,400 0,013 0,426 0,840 1,253 1,666 2,080 2,493 3,319
Probability 0,209 0,345 0,504 0,663 0,800 0,894 0,953 0,981 0,994 1,000
Table 7. Probabilities of the Project Completed Time is Less than X Values
18
6. CONCLUSION
Nowadays projects are more complex than being in the past and complexibility increases
uncertainty. In a conclusion deterministic approaches can’t afford to complex projects so
fuzzy numbers are more effective for high uncertainty situations. In the hospitals, time is the
most important issue so management department has to manage, control and plan the
projects with right numbers and fuzzy numbers provides to us to do that part right. In that
study, fuzzy numbers were explained and then, how you can use fuzzy numbers in a project
were explained with Fuzzy CPM (Critical Path Method) and Fuzzy PERT (Program Evaluation
and Review Technique). These tools provides to analyze flows and forecast project
completing time, and in healthcare sector, FCPM/FPERT provides to standardize care and
quality, reduce patient’s staying time. In addition, slack times between related activities can
be found with FCPM/FPERT tools and it provides to analyze space time and improve
productivity in the project.
19
8. REFERENCES
[ 1 ] G.S. Liang and T.C. Han, “Fuzzy Critical Path for Project Network”, Information
and Management Sciences, Vol. 15, pp. 29 – 40, 2004
[ 2 ] N.R. Shankar, V. Sireesha and P.P.B. Rao, “An Analytical Method for Finding
Critical Path in a Fuzzy Project Network”, Int. J. Contemp. Math. Sciences, Vol. 5., pp.
953 – 962, 2010
[ 3 ] N.R. Shankar, V. Sireesha, K.S. Rao and N. Vani, “Fuzzy Critical Path Method
Based on Metric Distance Ranking of Fuzzy Numbers”, Int. Journal Math. Analysis,
Vol. 4, pp. 995 – 1006, 2010
[ 4 ] M.F. El-Santawy and S.M. Abd-Allah, “The Longest Path Problem in Fuzzy
Project Networks : A Case Study”, Gen. Math. Notes, Vol.3, pp. 97 – 107, 2011
[ 5 ] V. Sireesha, K.S. Rao, N.R. Shankar and S.S. Babu, “Critical Path Analysis in the
Network with Fuzzy Interval Numbers as Activity Times”, International Journal of
Engineering Science and Technology (IJEST), Vol. 4, pp. 823 – 832, 2012
[ 6 ] L.C. Lee, S.C. Lee, K.Y. Wang and J.Y. Hsu, “Application of PERT/CPM to the
Care of Pulmonary Lobectomy Patients”, J Med Sci, Vol. 19, pp. 273 – 283, 1999
[ 7 ] V.R. Girija and M.S. Bhat, “Process Flow Analysis in the Emergency
Department of a Tertiary Care Hospital using Program Evaluation and Review
Technique (PERT)”, Journal of Health Management, Vol. 15, pp. 353 – 359, 2013
[ 8 ] U.K.M. Teichgräber, F. Neumann, J. Boeck, J. Ricke and R. Felix, “Process
Management in Computed Tomography : Using Critical Pathway Method to Design
and Improve Work Flow in Computed Tomography”, European Radiology, Vol.10, pp.
370 – 376, 2000
[ 9 ] L.A. Zadeh, “Fuzzy Sets a Basis for a Theory of Possibility”, Fuzzy Sets and
Systems, Vol. 1, pp. 3 – 28, 1978
[ 10 ] S.P. Chen and Y.J. Hsueh, “A Simple Approach to Fuzzy Critical Path Analysis in
Project Networks”, Applied Mathematical Modelling, Vol. 32, pp. 1289 – 1297, 2008
[ 11 ] Y. A. Ozcan, Quantitative Methods in Health Care Management: Techniques
and Applications, Ch. 13 : Project Management, Second Edition, Jossey-Bass, 2009
[ 12 ] M. Sharafi, F. Jolai, H. Iranmanesh and S.M. Hatefi, “A Model for Project
Scheduling with Fuzzy Precedence Links”, Australian Journal of Basic and Applied
Sciences, Vol. 2, pp. 1356-1361, 2008
[ 13 ] N.S. Pour, M. Kheranmand, M. Fallahc and S. Zeynali, “A New Method for
Critical Path Method with Fuzzy Processing Time”, Management Science Letters, Vol.
1, pp. 347–354, 2011
[ 14 ] Ö. Atlı and C. Kahraman, “Fuzzy Critical Path Method” , Journal of Engineering
and Natural Sciences – Sigma , Vol. 31 , pp. 128-140, 2013
[ 15 ] O. Moselhi and P. Lorterapong, “Fuzzy vs Probabilistic Scheduling”,
Automation and Robotics in Construction , Vol. 12 , pp. 441-448 ,1995