Embed Size (px)
Citation preview
IEEE
Proo
f
IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS, VOL. 00, NO. 00, 2014 1
Preventive Maintenance Prioritization Indexof Medical Equipment Using Quality
Function Deployment
1
2
3
Neven Saleh, Amr A. Sharawi, Manal Abd Elwahed, Alberto Petti, Daniele Puppato,and Gabriella Balestra
4
5
Abstract—Preventive maintenance is a core function of clinical6engineering, and it is essential to guarantee the correct functioning7of the equipment. The management and control of maintenance8activities are equally important to perform maintenance. As the9variety of medical equipment increases, accordingly the size of10maintenance activities increases, the need for better management11and control become essential. This paper aims to develop a new12model for preventive maintenance priority of medical equipment13using quality function deployment as a new concept in mainte-14nance of medical equipment. We developed a three-domain frame-15work model consisting of requirement, function, and concept. The16requirement domain is the house of quality matrix. The second17domain is the design matrix. Finally, the concept domain gener-18ates a prioritization index for preventive maintenance considering19the weights of critical criteria. According to the final scores of20those criteria, the prioritization action of medical equipment is21carried out. Our model proposes five levels of priority for preven-22tive maintenance. The model was tested on 200 pieces of medical23equipment belonging to 17 different departments of two hospi-24tals in Piedmont province, Italy. The dataset includes 70 different25types of equipment. The results show a high correlation between26risk-based criteria and the prioritization list.27
Index Terms—Medical equipment, preventive maintenance28(PM), prioritization, quality function deployment (QFD).29
I. INTRODUCTION30
PREVENTIVE maintenance (PM) is a core function of clin-31
ical engineering, having as objectives the assurance of32
ongoing safety and performance of medical devices and the33
preservation of the investment in the equipment through im-34
proved longevity [1]. PM is mainly a risk-based approach and35
considered as a core function of Clinical Engineering Depart-36
ment [2]. Despite its core role, the design and management of an37
effective PM program is not a simple matter. Adequate admin-38
istrative support is a requirement for an effective PM program.39
Manuscript received November 19, 2013; revised June 12, 2014; acceptedJuly 4, 2014. Date of publication; date of current version.
N. Saleh is with the Politecnico di Torino, 10129 Turin, Italy (e-mail:nevensaleh76@ gmail.com).
A. Sharawi and M. Abd Elwahed are with the Systems and BiomedicalEngineering Department, Faculty of Engineering, Cairo University, Giza, Egypt(e-mail: [email protected]; [email protected]).
A. Petti is with ASLBI, 13900 Biella, Italy (e-mail: [email protected]).
D. Puppato is with ARESS, 10122 Turin, Italy (e-mail: [email protected]).G. Balestra is with the Electronic and Telecommunication Department,
Politecnico di Torino, 10129 Turin, Italy (e-mail: [email protected]).Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.Digital Object Identifier 10.1109/JBHI.2014.2337895
The two key issues for PM are the procedures to be executed 40
and execution frequencies [1]–[3]. The procedures indicate the 41
necessary steps that are required to assure the performance of 42
the device, whereas the second key is the frequency at which 43
the set of procedures should be done. 44
Maintenance prioritization is a crucial task in management 45
systems, especially when there are more maintenance work or- 46
ders than available people or resources that can handle those 47
devices [4]. The literature is rich with different prioritization 48
approaches for medical equipment. In [5], Joseph and Mad- 49
hukumar have developed a model for PM index considering 50
risk level coefficient (RLC) of the instrument. RLC was cal- 51
culated through five different classified factors related to the 52
medical equipment electrical risk. The factors are static risk, 53
degree and quality of safety arrangements, insulation, physical 54
risk, and equipment contact with the patient. 55
In another work [6], the authors developed the system of infor- 56
mation technology and support system for maintenance actions 57
(SISMA). The system considers two aspects for PM evaluation: 58
technical and economic needs to assess PM plan for medical 59
equipment. Analytical Hierarchy Process was used in [7] as 60
a multicriteria decision making tool to develop a maintenance 61
priority index for medical equipment. 62
II. BACKGROUND 63
Quality function deployment (QFD) is one of the total qual- 64
ity management quantitative tools and techniques that could be 65
used to translate customer requirements and specifications into 66
appropriate technical or service requirements [8]. QFD was con- 67
ceived in Japan at the end of 1960s by Yoji Akao. The first usage 68
of QFD was implemented by Mizuno in 1972 to Mitsubishi’s 69
Kobe shipyard site [9]. 70
QFD uses visual matrices that link customer requirements, de- 71
sign requirements, target values, and competitive performance 72
into one chart [8]. Therefore, QFD is considered a quantitative 73
tool that facilitates evaluation of customer’s satisfaction. Typi- 74
cally, a QFD system can be broken into four interlinked phases 75
to fully deploy the customer needs phase by phase [10]. The 76
four-phase model consists of house of quality (HOQ) matrix, 77
design matrix, process planning matrix, and finally production 78
matrix [10], [11]. 79
Among the various matrices, the HOQ is commonly used in 80
the different applications. The HOQ matrix displays the voice of 81
customers (VOC) or customer requirements that are known as 82
2168-2194 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
IEEE
Proo
f
2 IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS, VOL. 00, NO. 00, 2014
Fig. 1. HOQ of function deployment [12].
WHATs against the technical requirements or voice of engineers83
(VOE) that are known as HOWs [8]–[11].84
In Fig. 1, a simplified matrix of HOQ is presented depicting85
the main parts of the matrix. The order suggested by letters86
A to F is normally followed during the process. Room “A”87
contains a list of customer needs, each of which is assessed88
against competitors, and the results which are absolute and rela-89
tive weights for customer needs prioritization are reported in to90
room “B.” Room “C” has the information necessary to transform91
the customer expectations into technical characteristics, and the92
correlation between each customer requirements and technical93
response is put into “D.” The roof, room “E,” considers the ex-94
tent to which the technical responses support each other. The95
prioritization of the technical characteristics, information on the96
competition, and technical targets weights all go into “F” [12].97
The most important parts in HOQ are “B” and “F,” respec-98
tively; more details can be found in [12]. The planning matrix99
“B” is calculated based on a comparison between the intended100
service within a hospital and other hospitals. In this matrix,101
the importance of customer requirements is evaluated, and the102
actual evaluations of the customer requirements are assigned.103
The goal is the expected value for each requirement, and then,104
the improvement ratio is calculated by dividing the goal to the105
evaluation of the intended requirements. The absolute weight106
is calculated by multiplication of goal and importance ratio,107
while the relative weight is the normalization of the absolute108
weight. Technical targets, room “F,” are prioritized through cal-109
culating the absolute weight of HOWs as given in (1) and then110
normalizing it to determine the relative weight [13]111
Absolute weight =∑
id WHAT XRWH (1)
where id WHAT is the importance degree of WHAT and RWH112
is the relationship value between WHAT and HOW.113
According to the characteristics of the QFD technique,114
our objective is to employee QFD as a new approach in115
medical equipment management to solve the problem of116
PM prioritization of medical equipment considering a set of117
influential criteria.118
Fig. 2. Proposed three-domain framework for PM of medical equipment pri-oritization.
III. METHODOLOGY 119
By using QFD, we proposed a three-domain framework for
Q1
120
PM priority, as illustrated in Fig. 3: requirement domain, func- 121
tion domain, and concept domain. 122
The first domain considers the customer requirements and the 123
technical characteristics that meet the customer requirements, 124
i.e., HOQ of the model. The second domain is the function 125
domain, in which the top technical criteria that resulted in first 126
domain will be measured through new criteria to identify the 127
critical criteria for PM prioritization, i.e., top HOWs of the first 128
domain becomes the new WHATs of the second domain. In 129
last domain, the concept domain, a priority score (PS) index is 130
generated considering the weights of critical criteria in order to 131
determine PM priority of medical equipment. 132
A. Requirement Domain 133
The HOQ of PM prioritization model is the requirement do- 134
main of this framework. First, customer needs and technical 135
characteristics should be identified. In general, the customers 136
of medical equipment in hospitals include all customers that 137
have a direct interface with medical equipment and who ex- 138
pect a range of services. In particular, the patients and clinical 139
staff are considered the customers (WHATs) of medical equip- 140
ment, whereas the Clinical Engineering Department is consid- 141
ered the one who is responsible to satisfy those requirements 142
(HOWs). 143
For patient’s requirements, no doubt that safety and avail- 144
ability of medical equipment are essential needs. Clinical staff 145
requirements were chosen based upon the literature [14] and 146
experience. Table I depicts the proposed customer’s require- 147
ments and technical characteristics. Prioritization of customer’s 148
requirements is performed considering a comparison between 149
Italian hospital and Jordanian hospital [14]. 150
Table I shows that customer’s requirements are met by ad- 151
dressing five technical characteristics: risk, performance assur- 152
ance, user competence, costs, and standard compliance, which 153
are presented with its subcriteria in this table. The requirement 154
domain (HOQ) is shown in Fig. 3. 155
The matrix is organized referring to Fig. 1; the left column 156
contains customer requirements (VOC), whereas the main part 157
IEEE
Proo
f
SALEH et al.: PM PRIORITIZATION INDEX OF MEDICAL EQUIPMENT USING QFD 3
TABLE ICUSTOMER REQUIREMENTS AND TECHNICAL CHARACTERISTICS OF HOQ
OF THE PROPOSED FRAMEWORK
Customer Requirements Technical Requirements
Criteria Subcriteria
Safety of medical.equipment.Efficiency.Durability.Quick response oftechnical team.Back up availability.Check the device aftermaintenance.Regular monitoring ofthe devices.
Risk Physical risk FunctionMaintenancerequirements
Priority based onimportance.Obvious operatinginstructions.Knowledge ofmaintained devices.Existence of a contactperson 24 h.Avoiding suspension ofservices.
PerformanceAssurance
Mission criticalityFunction verification AgeLabeling Electrical safety
Parts replacementRegular inspection
UserCompetence
Qualification oftechnicians Complexity
of devices Equippedworkshop Test equipment
availability Servicemanual availabilityActivities recording
Costs Updating or loan Spareparts availability Service
provider typeStandards Meet standards
of the matrix contains the technical characteristics (VOE) seg-158
mented into five columns and the relationship matrix. The right159
column is the planning matrix of HOQ. The bottom room is160
the technical target matrix. The relationships between WHATs161
and HOWs are indicated by scores 9 for strong relation, 3162
for medium relation, 1 for low relation, and blank for no163
relation [11]–[13].164
In order to demonstrate how customer needs are prioritized,165
consider “safety” requirement as an example, by using five-166
point scale, we evaluate 5 for importance, 3 for Italian hospital,167
5 for Jordanian hospital, and 5 for goal. Regarding our per-168
spective to Italian hospital, the improvement ratio is calculated169
by dividing the goal to Italian hospital, i.e., 5/3. The absolute170
weight is calculated through multiplication of improvement ra-171
tio by importance, i.e., 1.7 × 5; meanwhile, the relative weight172
is obtained by normalizing the absolute weight, i.e., (8.3/91.5)173
∗ 100. As such for technical targets, to explain how technical174
criteria priority is identified, we consider “physical risk” as an175
example. Utilizing formula (1), the absolute weight of “physical176
risk” equals 9 × 9.1 + 9 × 7.3 + 9 × 4.4 + 3 × 8.7 = 213.3,177
and the relative weight is calculated also by normalization178
to become 5.03.179
B. Function Domain180
The next stage of our proposed model is to identify the critical181
criteria among the technical criteria for PM priority. We selected182
the top 11 criteria of technical terms based upon their weights 183
and importance to become the inputs (WHATs) of the second 184
matrix as illustrated in Fig. 4. The design matrix is constructed 185
with the same way that is followed in building the HOQ in 186
Fig. 3, except the planning matrix; it is the top 11 output relative 187
weights of HOQ. 188
For better representation of five addressing criteria dimen- 189
sions, we propose all subcriteria with relative weight greater 190
than 4.5% to be selected as top criteria in order to cover a wide 191
range of criteria ranging from risk criteria with its clear impact 192
in PM to a regular inspection since it is not regularly followed by 193
a lot of hospitals especially in developing countries. The criteria 194
are function, mission criticality, service provider type, standards 195
compliance, maintenance requirements, age, functional verifi- 196
cations, team qualification, device complexity, physical risk, and 197
regular inspection. 198
We classified the critical criteria into three categories for 199
HOWs of the second matrix: risk-based criteria including func- 200
tion, physical risk, and maintenance requirements; mission- 201
based criteria including utilization level, area criticality, and 202
device criticality; and finally maintenance-based criteria incor- 203
porating, failure rate, useful life ratio, device complexity, num- 204
ber of missed maintenance, and downtime ratio. The criteria 205
were selected based upon the literature [3], [5]–[7], [14], in 206
addition to the authors experience. 207
The matrix roof depicts the relations among HOWs [15]. 208
Strong relation is indicated by •, medium by �, and low by 209
o. The relationships are proposed according to author’s expe- 210
rience. As shown in Fig. 4, the planning part of design matrix 211
(importance of WHATs) is the resultant relative weight of top 212
11 criteria of first domain. The critical targets of the technical 213
criteria are considered based upon the proposed thresholds as 214
presented in Table II. On the other hand, the technical targets, 215
i.e., the absolute weight and relative weight of the technical 216
criteria are calculated as the same for requirement domain. In 217
fact, the resultant ranking of most critical criteria for PM prior- 218
itization point to risk-based criteria is topping the list of criteria 219
(44.9%) that it is logical with previous survey, followed by 220
mission-based and maintenance-based. 221
C. Concept Domain 222
The concept domain is the output of the design matrix. The 223
output is a prioritization equation considering the 11 most crit- 224
ical factors with the resultant weights. Equation (2) generates 225
the priority index of PM that is presented as scores. Table II 226
illustrates a brief description of the critical criteria and their 227
proposed scores. 228
PS = 11.7(FN) + 12.8(PR) + 20.4(MR) + 11(UL)
+ 6.5(AC) + 11.4(DC) + 8.3(FR) + 5.1(LR)
+ 6.3(CM) + 3.4(MM) + 3.1(DR) (2)
where 229
PS priority score; 230
FN function of equipment; 231
PR physical risk; 232
MR maintenance requirements; 233
IEEE
Proo
f
4 IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS, VOL. 00, NO. 00, 2014
Fig. 3. HOQ matrix of the QFD model for PM prioritization of medical equipment (requirement domain).
Fig. 4. Design matrix of the QFD model for pm prioritization of medical equipment (function domain).
IEEE
Proo
f
SALEH et al.: PM PRIORITIZATION INDEX OF MEDICAL EQUIPMENT USING QFD 5
TABLE IIBRIEF DESCRIPTION OF CRITICAL PARAMETERS AND THEIR PROPOSED SCORES
Parameter Description Thresholds Scores
Function Device function Life supportTherapeuticDiagnostic /monitoringAnalytical
Miscellaneous
543
21
Physical risk Probable harmscaused byequipment
failure
DeathInjury
MisdiagnosisEquipment
damageNo risk
5432
1Maintenancerequirements
Maintenanceactivities
depending onequipment type
ExtensiveA above average
AverageBelow average
Minimal
54321
Utilization level Number ofworking days a
week
4 days3/4 days<3 days
321
Area criticality Assessment ofarea criticality
for patients
UrgentIntensity care
unitsDiagnostic areaLaw intensity
areaNon clinical area
54
32
1Device criticality The importance
level ofequipment inserviced area
CriticalImportantNecessary
321
Failure rate Number offailures a year
based on devicecriticality level
�2 for critical,�4 for
important, �5for necessary
1 for critical, 2/3for important,
3/4 for necessary0 for critical, �1
for important,�2 for necessary
3
2
1
Useful life ratio Ratio betweenage to expected
life time of adevice
Ratio > 80 %50% < Ratio
�80%Ratio � 50 %
32
1Device complexity Technical
complexitybased on a
model
Score 6—8Score 3—5Score 0—2
321
Missed maintenance Number ofmissed
maintenance ayear
� 210
321
Downtime ratio Ratio betweenthe duration ofdowntime in
days to days ayear
Ratio � 20 %10% � Ratio <
20%Ratio < 10 %
32
1
UL utilization level;234
AC area criticality;235
DC device criticality;236
FR failure rate;237
LR useful life ratio;238
CM device complexity;239
MM missed maintenance;240
DR downtime ratio.241
TABLE IIIDATA SAMPLE OF INVESTIGATED EQUIPMENT FOR PM PRIORITY
Equipment FN PR MR UL AC DC FR LR CM MM DR PS PS %
Ventilator 5 5 5 2 5 3 2 3 3 3 1 377 93C-arm 3 3 4 2 5 3 3 3 3 2 2 316 78Monitor 3 3 3 3 3 2 3 3 2 1 1 269 66Centrifuge 2 3 3 3 2 2 1 3 1 1 1 228 56Scale 1 1 1 1 3 1 1 2 1 2 1 121 30
Fig. 5. PM priority index groups based on PS value of the proposed QFDmodel.
The complexity model [16] assesses the technical complexity 242
of a device considering four factors: equipment maintainability, 243
installation requirements, repair, and connectivity. The scores 244
are given for each factor range from 0 to 2 for evaluation based 245
upon the complexity level for each device in the list. 246
IV. RESULTS 247
For model verification, we utilized a dataset of 200 medical 248
equipment of two hospitals with one management system in 249
Piedmont province, Italy. Seventy different types of equipment 250
belonging to 17 different kinds of departments were analyzed 251
along data collected for one year (2012). Table III shows a 252
sample data of various types of investigated equipment along 253
with PS based upon our proposed model. 254
The scores in Table III are obtained referring to the scores that 255
are given in Table II. For instance, considering the “Ventilator” 256
device, and according to concept domain, “Function” is life sup- 257
port, “Physical Risk” is death, “Maintenance Requirements” is 258
extensive, “Area Criticality” is urgent, and “Device Criticality” 259
is critical; meanwhile, other parameters “Failure Rate,” “Useful 260
Life,” “Missed Maintenance,” and “Downtime Ratio” are de- 261
pending on the actual status of each device. Device complexity 262
is calculated relied on complexity model [16]. Accordingly, by 263
utilizing formula (2), the PS for every device is calculated as 264
shown in Table III. 265
By using the result of PS percentages, the PM priority is 266
classified into five categories as illustrated in Fig. 5. The first 267
class is very high-priority class and includes equipment that 268
supposed to be maintained within two weeks with PS percentage 269
equal or greater than 80. In second class, high-priority PM 270
should be performed within one month if priority percentage 271
in range 70–80. Class 3 is medium priority and contains all 272
IEEE
Proo
f
6 IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS, VOL. 00, NO. 00, 2014
Fig. 6. Results of the PM prioritization index for investigated medical equip-ment.
equipment that should be considered for PM within two months273
in case of priority percentage in range 60–70.274
Class 4 is low priority and includes all equipment with pri-275
ority percentage of 50 to 60, and in this case, PM should be276
performed within three months. Finally, all equipment with pri-277
ority percentage less than 50 could be visually inspected and278
considered for next PM as minimal PM.279
In our dataset of equipment and according to the proposed280
model and priority classification, the results indicate that 30281
devices (15%) need very high priority PM, 39 devices (19%)282
should be included as high priority, 59 devices (30%) should be283
considered for medium priority, 54 devices (27%) for low prior-284
ity, and finally, 18 devices (9%) should be with minimal priority285
PM. Fig. 6 presents dataset output classification considering the286
QFD model.287
By analyzing the results, very high priority class incor-288
porates all equipment with high-risk criteria, relatively high-289
mission-based criteria, and also with high-complexity level.290
High-priority class contains relatively high-risk criteria and291
mission-based criteria in addition to high missed maintenance.292
Medium priority class is considered for equipment with rela-293
tively high utilization level, area criticality, and old equipment.294
Low-priority class contains old not risky devices. Relatively sta-295
ble equipment does not need PM. The results are consistent with296
the classification given by an experienced clinical engineer. In297
other words, the devices that pose risk for patients and users in298
case of PM omission, old devices, and complex devices such as299
radiology equipment are highly considered for PM; meanwhile,300
low risk devices, reliable devices, and relatively new devices are301
modestly considered for PM. Moreover, as indicated in Fig. 5,302
the risk-based criteria contribute to approximately 45% of the303
weight of priority index. In fact, this contribution reflects high304
correlation existence between risk assessment and PM manage-305
ment of medical equipment.306
V. CONCLUSION307
In this study, QFD is presented for the first time to solve the308
problem of PM prioritization of medical equipment. The pro-309
posed model has proven its validity in real environment correctly310
separating equipment that needs PM from those that do not need311
it. The model was tested based upon the periodical schedule of312
the hospitals. It is important to note that the classification is 313
founded considering the requirements of patients and clinical 314
staff. By analyzing the results, we can state that the risk-based 315
criteria have a great impact on PM prioritization decision in ad- 316
dition to criticality and age of medical equipment. Also, the work 317
highlights the importance of existence of a detailed history for 318
every device that helps the decision makers to manage the med- 319
ical equipment obviously. The model can be used every time a 320
PM is to be planned modifying only the instruments data. More- 321
over, the model development could be improved by addressing 322
the customer requirements according to Kano’s model to clar- 323
ify the attitudes of customer satisfactions in medical equipment 324
PM. In addition, the developed QFD model can be implemented 325
in other stages of management of medical equipment such as 326
acquisition and procurement of medical equipment. The model 327
could be extended to maintenance agencies in order to organize 328
their PM programs. 329
REFERENCES 330
[1] W. A. Hyman, “The Theory and practice of preventive maintenance,” 331J. Clin. Eng., vol. 28, no. 1, pp. 31–36. Jan.–Mar. 2003. 332
[2] M. Ridgway, “Optimizing our PM programs,” Biomed. Instrum. Technol., 333vol. 43, no. 3, pp. 244–254. May/Jun. 2009. 334
[3] R. B. Arslan and Y. Ulgen, “Smart IPM: An adaptive tool for the preventive 335maintenance of medical equipment,” in Proc. IEEE 23rd Annu. Int. Conf. 336Eng. Med. Biol. Soc., Istanbul, Turkey, 2001, vol. 4, pp. 3950–3953. 337
[4] J. Ni and X. Jin, “Decision support systems for effective maintenance 338operations,” CIRP Ann.—Manuf. Technol., vol. 61, pp. 411–414, 2012. 339
[5] J. Joseph and S. Madhukumar, “A novel approach to data driven preventive 340maintenance scheduling of medical instruments,” in Proc. Int. Conf. Syst. 341Med. Biol., Kharagpur, India, 2010, pp. 193–197. 342
[6] R. Miniati, F. Dori, and G. B. Gentili, “Design of a decision support system 343for preventive maintenance planning in health structures,” Technol. Health 344Care J., vol. 20, pp. 205–214, 2012. 345
[7] S. Taghipour, D. Banjevic, and A. K. S. Jardine, “Prioritization of medical 346equipment for maintenance decisions,” J. Oper. Res. Soc., vol. 62, no. 9, 347pp. 1666–1687, 2010. 348
[8] S. O. Duffuaa, A. H. Al-Ghamdi, and A. Al-Amer, “Quality function 349deployment in maintenance work planning process,” in Proc. 6th Saudi 350Eng. Conf., 2002, vol. 4, pp. 503–512. 351
[9] B. M. Deros, N, Rahman, M. N. A. Rahman, A. R. Ismail, and 352A. H. Said, “Application of quality function deployment to study criti- 353cal service quality characteristics and performance measures,” Eur. J. Sci. 354Res., vol. 33, no. 3, pp. 398–410, 2009. 355
[10] S. Bennur and B. Jin, “A conceptual process of implementing quality 356apparel retail store attributes: An application of Kano’s model and the 357quality function deployment approach,” Int. J. Bus., Humanities Technol., 358vol. 2, no. 1, pp. 174–183, 2012 359
[11] X. X. Shen, K. C. Tan, and M. Xie, “An integrated approach to innovative 360product development using Kano’s model and QFD,” Eur. J. Innovative 361Manag., vol. 3, no. 2, pp. 91–99, 2000. 362
[12] D. J. Delgado, K. E. Bampton, and E. Aspinwall, “Quality function 363deployment in construction,” Constr. Manag. Econ. J., vol. 24, no. 6, 364pp. 597–609, 2007. 365
[13] L.K. Chan and M.-L. Wu, “A systematic approach to quality function 366deployment with a full illustrative example,” Int. J. Manag. Sci., vol. 33, 367no. 2, pp. 119–139, 2005. 368
[14] A. Al-Bashir, M. Al-Rawashdeh, R. Al-Hadithi, A. Al-Ghandoor, and M. 369Barghash, “Building medical devices maintenance system through quality 370function deployment,” Jordan J. Mech. Ind. Eng., vol. 25, no. 1, pp. 25–36, 371Feb. 2012. 372
[15] N. Z. Haron and F. L. M. Kairudin, “The application of quality function 373deployment (QFD) in the design phase of industrialized building system 374(IBS) apartment construction project,” Eur. Int. J. Sci. Technol., vol. 1, 375no. 3, pp. 56–66, 2012. 376
[16] N. F. Youssef and W. A. Hyman, “A medical device complexity model: A 377new approach to medical equipment management,” J. Clin. Eng., vol. 34, 378no. 2, pp. 94–98, 2009. 379
IEEE
Proo
f
SALEH et al.: PM PRIORITIZATION INDEX OF MEDICAL EQUIPMENT USING QFD 7
Neven Saleh received the B.Sc. degree in electronics and electrical commu-380nication and the diploma degree in electrical communication from Mansoura381University, Mansoura, Egypt, in 1999 and 2001, respectively, and the M.Sc.382degree in systems and biomedical engineering from Cairo University, Giza,383Egypt, in 2011. She is currently working toward the Ph.D. degree in Biomedical384Engineering from Cairo University, and also from the Politecnico di Torino,385Turin, Italy, since she was granted a channel scholarship.386
Since 2001, she has been a Clinical Engineer with Mansoura University387Children Hospital, Mansoura.388
389
Amr A. Sharawi received the B.Sc. degree in electronics and communication390and the M.Sc. and Ph.D. degrees in systems and biomedical engineering, all391from Cairo University, Giza, Egypt, in 1976, 1981, and 1991, respectively.392
Since 2001, he has been an Associate Professor in biomedical engineering,393Faculty of Engineering, Cairo University.394
395
Manal Abd Elwahed received the B.Sc., M.Sc., and Ph.D. degrees in biomedi-396cal engineering from the Faculty of Engineering, Cairo University, Giza, Egypt,397in 1987, 1992, and1999, respectively.398
She is currently an Associate Professor in the Systems and Biomedical En-399gineering Department, Faculty of Engineering, Cairo University.400
401
Alberto Petti received the first M.Sc. degree in biomedical engineering and 402the second M.Sc. degree in management and health technologies from the 403Politecnico di Torrino, Torino, Italy, in 2007 and 2008, respectively. 404
Since 2000, he has been responsible for the clinical engineering service in 405the Public Hospital of Biella, Italy. He is currently involved in realization of the 406new hospital of the city. He is currently a Mechanical Engineer. His research 407interests include the applications of methods for better organizations of the 408public health system, involving technologies and ICT tools, particularly related 409to management of health technologies. 410
411
Daniele Puppato received the B.Sc. and M.Sc. degrees in biomedical engineer- 412ing from the Politecnico di Torrino, Torino, Italy, in 2003 and 2005, respectively. 413
From 2006 to June 2013, he was with AReSS Piemonte as a Health Care 414Technology Management Leader. He is currently working as a Consultant in a 415leading Italian company developing software solutions for technical manage- 416ment of hospitals. 417
Mr. Puppato is a Member of the Italian Association of Clinical Engineers. 418419
Gabriella Balestra received the M.Sc. degree in computer science from the 420University of Turin, Turin, Italy, in 1980, and the Ph.D. degree in computer and 421system engineering from the Politecnico di Torino, Turin, in 1989. 422
She is currently an Assistant Professor in the Department of Electronics 423and Telecommunications, Politecnico di Torino, where she is a Lecturer for 424the Biomedical Engineering course. She also teaches decision support systems, 425biomedical data classification techniques, and medical informatics. 426
Dr. Balestra is a member of GNB, the Italian National Group of Bioengi- 427neering, and the Association for the Advancement of Medical Instrumentation. Q2428
429
IEEE
Proo
f
QUERIES430
Q1. Author: Fig. 2 is not cited in the text. Please cite it at an appropriate place.431
Q2. Author: Please provide the expansion of “GNB.”432
IEEE
Proo
f
IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS, VOL. 00, NO. 00, 2014 1
Preventive Maintenance Prioritization Indexof Medical Equipment Using Quality
Function Deployment
1
2
3
Neven Saleh, Amr A. Sharawi, Manal Abd Elwahed, Alberto Petti, Daniele Puppato,and Gabriella Balestra
4
5
Abstract—Preventive maintenance is a core function of clinical6engineering, and it is essential to guarantee the correct functioning7of the equipment. The management and control of maintenance8activities are equally important to perform maintenance. As the9variety of medical equipment increases, accordingly the size of10maintenance activities increases, the need for better management11and control become essential. This paper aims to develop a new12model for preventive maintenance priority of medical equipment13using quality function deployment as a new concept in mainte-14nance of medical equipment. We developed a three-domain frame-15work model consisting of requirement, function, and concept. The16requirement domain is the house of quality matrix. The second17domain is the design matrix. Finally, the concept domain gener-18ates a prioritization index for preventive maintenance considering19the weights of critical criteria. According to the final scores of20those criteria, the prioritization action of medical equipment is21carried out. Our model proposes five levels of priority for preven-22tive maintenance. The model was tested on 200 pieces of medical23equipment belonging to 17 different departments of two hospi-24tals in Piedmont province, Italy. The dataset includes 70 different25types of equipment. The results show a high correlation between26risk-based criteria and the prioritization list.27
Index Terms—Medical equipment, preventive maintenance28(PM), prioritization, quality function deployment (QFD).29
I. INTRODUCTION30
PREVENTIVE maintenance (PM) is a core function of clin-31
ical engineering, having as objectives the assurance of32
ongoing safety and performance of medical devices and the33
preservation of the investment in the equipment through im-34
proved longevity [1]. PM is mainly a risk-based approach and35
considered as a core function of Clinical Engineering Depart-36
ment [2]. Despite its core role, the design and management of an37
effective PM program is not a simple matter. Adequate admin-38
istrative support is a requirement for an effective PM program.39
Manuscript received November 19, 2013; revised June 12, 2014; acceptedJuly 4, 2014. Date of publication; date of current version.
N. Saleh is with the Politecnico di Torino, 10129 Turin, Italy (e-mail:nevensaleh76@ gmail.com).
A. Sharawi and M. Abd Elwahed are with the Systems and BiomedicalEngineering Department, Faculty of Engineering, Cairo University, Giza, Egypt(e-mail: [email protected]; [email protected]).
A. Petti is with ASLBI, 13900 Biella, Italy (e-mail: [email protected]).
D. Puppato is with ARESS, 10122 Turin, Italy (e-mail: [email protected]).G. Balestra is with the Electronic and Telecommunication Department,
Politecnico di Torino, 10129 Turin, Italy (e-mail: [email protected]).Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.Digital Object Identifier 10.1109/JBHI.2014.2337895
The two key issues for PM are the procedures to be executed 40
and execution frequencies [1]–[3]. The procedures indicate the 41
necessary steps that are required to assure the performance of 42
the device, whereas the second key is the frequency at which 43
the set of procedures should be done. 44
Maintenance prioritization is a crucial task in management 45
systems, especially when there are more maintenance work or- 46
ders than available people or resources that can handle those 47
devices [4]. The literature is rich with different prioritization 48
approaches for medical equipment. In [5], Joseph and Mad- 49
hukumar have developed a model for PM index considering 50
risk level coefficient (RLC) of the instrument. RLC was cal- 51
culated through five different classified factors related to the 52
medical equipment electrical risk. The factors are static risk, 53
degree and quality of safety arrangements, insulation, physical 54
risk, and equipment contact with the patient. 55
In another work [6], the authors developed the system of infor- 56
mation technology and support system for maintenance actions 57
(SISMA). The system considers two aspects for PM evaluation: 58
technical and economic needs to assess PM plan for medical 59
equipment. Analytical Hierarchy Process was used in [7] as 60
a multicriteria decision making tool to develop a maintenance 61
priority index for medical equipment. 62
II. BACKGROUND 63
Quality function deployment (QFD) is one of the total qual- 64
ity management quantitative tools and techniques that could be 65
used to translate customer requirements and specifications into 66
appropriate technical or service requirements [8]. QFD was con- 67
ceived in Japan at the end of 1960s by Yoji Akao. The first usage 68
of QFD was implemented by Mizuno in 1972 to Mitsubishi’s 69
Kobe shipyard site [9]. 70
QFD uses visual matrices that link customer requirements, de- 71
sign requirements, target values, and competitive performance 72
into one chart [8]. Therefore, QFD is considered a quantitative 73
tool that facilitates evaluation of customer’s satisfaction. Typi- 74
cally, a QFD system can be broken into four interlinked phases 75
to fully deploy the customer needs phase by phase [10]. The 76
four-phase model consists of house of quality (HOQ) matrix, 77
design matrix, process planning matrix, and finally production 78
matrix [10], [11]. 79
Among the various matrices, the HOQ is commonly used in 80
the different applications. The HOQ matrix displays the voice of 81
customers (VOC) or customer requirements that are known as 82
2168-2194 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
IEEE
Proo
f
2 IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS, VOL. 00, NO. 00, 2014
Fig. 1. HOQ of function deployment [12].
WHATs against the technical requirements or voice of engineers83
(VOE) that are known as HOWs [8]–[11].84
In Fig. 1, a simplified matrix of HOQ is presented depicting85
the main parts of the matrix. The order suggested by letters86
A to F is normally followed during the process. Room “A”87
contains a list of customer needs, each of which is assessed88
against competitors, and the results which are absolute and rela-89
tive weights for customer needs prioritization are reported in to90
room “B.” Room “C” has the information necessary to transform91
the customer expectations into technical characteristics, and the92
correlation between each customer requirements and technical93
response is put into “D.” The roof, room “E,” considers the ex-94
tent to which the technical responses support each other. The95
prioritization of the technical characteristics, information on the96
competition, and technical targets weights all go into “F” [12].97
The most important parts in HOQ are “B” and “F,” respec-98
tively; more details can be found in [12]. The planning matrix99
“B” is calculated based on a comparison between the intended100
service within a hospital and other hospitals. In this matrix,101
the importance of customer requirements is evaluated, and the102
actual evaluations of the customer requirements are assigned.103
The goal is the expected value for each requirement, and then,104
the improvement ratio is calculated by dividing the goal to the105
evaluation of the intended requirements. The absolute weight106
is calculated by multiplication of goal and importance ratio,107
while the relative weight is the normalization of the absolute108
weight. Technical targets, room “F,” are prioritized through cal-109
culating the absolute weight of HOWs as given in (1) and then110
normalizing it to determine the relative weight [13]111
Absolute weight =∑
id WHAT XRWH (1)
where id WHAT is the importance degree of WHAT and RWH112
is the relationship value between WHAT and HOW.113
According to the characteristics of the QFD technique,114
our objective is to employee QFD as a new approach in115
medical equipment management to solve the problem of116
PM prioritization of medical equipment considering a set of117
influential criteria.118
Fig. 2. Proposed three-domain framework for PM of medical equipment pri-oritization.
III. METHODOLOGY 119
By using QFD, we proposed a three-domain framework for
Q1
120
PM priority, as illustrated in Fig. 3: requirement domain, func- 121
tion domain, and concept domain. 122
The first domain considers the customer requirements and the 123
technical characteristics that meet the customer requirements, 124
i.e., HOQ of the model. The second domain is the function 125
domain, in which the top technical criteria that resulted in first 126
domain will be measured through new criteria to identify the 127
critical criteria for PM prioritization, i.e., top HOWs of the first 128
domain becomes the new WHATs of the second domain. In 129
last domain, the concept domain, a priority score (PS) index is 130
generated considering the weights of critical criteria in order to 131
determine PM priority of medical equipment. 132
A. Requirement Domain 133
The HOQ of PM prioritization model is the requirement do- 134
main of this framework. First, customer needs and technical 135
characteristics should be identified. In general, the customers 136
of medical equipment in hospitals include all customers that 137
have a direct interface with medical equipment and who ex- 138
pect a range of services. In particular, the patients and clinical 139
staff are considered the customers (WHATs) of medical equip- 140
ment, whereas the Clinical Engineering Department is consid- 141
ered the one who is responsible to satisfy those requirements 142
(HOWs). 143
For patient’s requirements, no doubt that safety and avail- 144
ability of medical equipment are essential needs. Clinical staff 145
requirements were chosen based upon the literature [14] and 146
experience. Table I depicts the proposed customer’s require- 147
ments and technical characteristics. Prioritization of customer’s 148
requirements is performed considering a comparison between 149
Italian hospital and Jordanian hospital [14]. 150
Table I shows that customer’s requirements are met by ad- 151
dressing five technical characteristics: risk, performance assur- 152
ance, user competence, costs, and standard compliance, which 153
are presented with its subcriteria in this table. The requirement 154
domain (HOQ) is shown in Fig. 3. 155
The matrix is organized referring to Fig. 1; the left column 156
contains customer requirements (VOC), whereas the main part 157
IEEE
Proo
f
SALEH et al.: PM PRIORITIZATION INDEX OF MEDICAL EQUIPMENT USING QFD 3
TABLE ICUSTOMER REQUIREMENTS AND TECHNICAL CHARACTERISTICS OF HOQ
OF THE PROPOSED FRAMEWORK
Customer Requirements Technical Requirements
Criteria Subcriteria
Safety of medical.equipment.Efficiency.Durability.Quick response oftechnical team.Back up availability.Check the device aftermaintenance.Regular monitoring ofthe devices.
Risk Physical risk FunctionMaintenancerequirements
Priority based onimportance.Obvious operatinginstructions.Knowledge ofmaintained devices.Existence of a contactperson 24 h.Avoiding suspension ofservices.
PerformanceAssurance
Mission criticalityFunction verification AgeLabeling Electrical safety
Parts replacementRegular inspection
UserCompetence
Qualification oftechnicians Complexity
of devices Equippedworkshop Test equipment
availability Servicemanual availabilityActivities recording
Costs Updating or loan Spareparts availability Service
provider typeStandards Meet standards
of the matrix contains the technical characteristics (VOE) seg-158
mented into five columns and the relationship matrix. The right159
column is the planning matrix of HOQ. The bottom room is160
the technical target matrix. The relationships between WHATs161
and HOWs are indicated by scores 9 for strong relation, 3162
for medium relation, 1 for low relation, and blank for no163
relation [11]–[13].164
In order to demonstrate how customer needs are prioritized,165
consider “safety” requirement as an example, by using five-166
point scale, we evaluate 5 for importance, 3 for Italian hospital,167
5 for Jordanian hospital, and 5 for goal. Regarding our per-168
spective to Italian hospital, the improvement ratio is calculated169
by dividing the goal to Italian hospital, i.e., 5/3. The absolute170
weight is calculated through multiplication of improvement ra-171
tio by importance, i.e., 1.7 × 5; meanwhile, the relative weight172
is obtained by normalizing the absolute weight, i.e., (8.3/91.5)173
∗ 100. As such for technical targets, to explain how technical174
criteria priority is identified, we consider “physical risk” as an175
example. Utilizing formula (1), the absolute weight of “physical176
risk” equals 9 × 9.1 + 9 × 7.3 + 9 × 4.4 + 3 × 8.7 = 213.3,177
and the relative weight is calculated also by normalization178
to become 5.03.179
B. Function Domain180
The next stage of our proposed model is to identify the critical181
criteria among the technical criteria for PM priority. We selected182
the top 11 criteria of technical terms based upon their weights 183
and importance to become the inputs (WHATs) of the second 184
matrix as illustrated in Fig. 4. The design matrix is constructed 185
with the same way that is followed in building the HOQ in 186
Fig. 3, except the planning matrix; it is the top 11 output relative 187
weights of HOQ. 188
For better representation of five addressing criteria dimen- 189
sions, we propose all subcriteria with relative weight greater 190
than 4.5% to be selected as top criteria in order to cover a wide 191
range of criteria ranging from risk criteria with its clear impact 192
in PM to a regular inspection since it is not regularly followed by 193
a lot of hospitals especially in developing countries. The criteria 194
are function, mission criticality, service provider type, standards 195
compliance, maintenance requirements, age, functional verifi- 196
cations, team qualification, device complexity, physical risk, and 197
regular inspection. 198
We classified the critical criteria into three categories for 199
HOWs of the second matrix: risk-based criteria including func- 200
tion, physical risk, and maintenance requirements; mission- 201
based criteria including utilization level, area criticality, and 202
device criticality; and finally maintenance-based criteria incor- 203
porating, failure rate, useful life ratio, device complexity, num- 204
ber of missed maintenance, and downtime ratio. The criteria 205
were selected based upon the literature [3], [5]–[7], [14], in 206
addition to the authors experience. 207
The matrix roof depicts the relations among HOWs [15]. 208
Strong relation is indicated by •, medium by �, and low by 209
o. The relationships are proposed according to author’s expe- 210
rience. As shown in Fig. 4, the planning part of design matrix 211
(importance of WHATs) is the resultant relative weight of top 212
11 criteria of first domain. The critical targets of the technical 213
criteria are considered based upon the proposed thresholds as 214
presented in Table II. On the other hand, the technical targets, 215
i.e., the absolute weight and relative weight of the technical 216
criteria are calculated as the same for requirement domain. In 217
fact, the resultant ranking of most critical criteria for PM prior- 218
itization point to risk-based criteria is topping the list of criteria 219
(44.9%) that it is logical with previous survey, followed by 220
mission-based and maintenance-based. 221
C. Concept Domain 222
The concept domain is the output of the design matrix. The 223
output is a prioritization equation considering the 11 most crit- 224
ical factors with the resultant weights. Equation (2) generates 225
the priority index of PM that is presented as scores. Table II 226
illustrates a brief description of the critical criteria and their 227
proposed scores. 228
PS = 11.7(FN) + 12.8(PR) + 20.4(MR) + 11(UL)
+ 6.5(AC) + 11.4(DC) + 8.3(FR) + 5.1(LR)
+ 6.3(CM) + 3.4(MM) + 3.1(DR) (2)
where 229
PS priority score; 230
FN function of equipment; 231
PR physical risk; 232
MR maintenance requirements; 233
IEEE
Proo
f
4 IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS, VOL. 00, NO. 00, 2014
Fig. 3. HOQ matrix of the QFD model for PM prioritization of medical equipment (requirement domain).
Fig. 4. Design matrix of the QFD model for pm prioritization of medical equipment (function domain).
IEEE
Proo
f
SALEH et al.: PM PRIORITIZATION INDEX OF MEDICAL EQUIPMENT USING QFD 5
TABLE IIBRIEF DESCRIPTION OF CRITICAL PARAMETERS AND THEIR PROPOSED SCORES
Parameter Description Thresholds Scores
Function Device function Life supportTherapeuticDiagnostic /monitoringAnalytical
Miscellaneous
543
21
Physical risk Probable harmscaused byequipment
failure
DeathInjury
MisdiagnosisEquipment
damageNo risk
5432
1Maintenancerequirements
Maintenanceactivities
depending onequipment type
ExtensiveA above average
AverageBelow average
Minimal
54321
Utilization level Number ofworking days a
week
4 days3/4 days<3 days
321
Area criticality Assessment ofarea criticality
for patients
UrgentIntensity care
unitsDiagnostic areaLaw intensity
areaNon clinical area
54
32
1Device criticality The importance
level ofequipment inserviced area
CriticalImportantNecessary
321
Failure rate Number offailures a year
based on devicecriticality level
�2 for critical,�4 for
important, �5for necessary
1 for critical, 2/3for important,
3/4 for necessary0 for critical, �1
for important,�2 for necessary
3
2
1
Useful life ratio Ratio betweenage to expected
life time of adevice
Ratio > 80 %50% < Ratio
�80%Ratio � 50 %
32
1Device complexity Technical
complexitybased on a
model
Score 6—8Score 3—5Score 0—2
321
Missed maintenance Number ofmissed
maintenance ayear
� 210
321
Downtime ratio Ratio betweenthe duration ofdowntime in
days to days ayear
Ratio � 20 %10% � Ratio <
20%Ratio < 10 %
32
1
UL utilization level;234
AC area criticality;235
DC device criticality;236
FR failure rate;237
LR useful life ratio;238
CM device complexity;239
MM missed maintenance;240
DR downtime ratio.241
TABLE IIIDATA SAMPLE OF INVESTIGATED EQUIPMENT FOR PM PRIORITY
Equipment FN PR MR UL AC DC FR LR CM MM DR PS PS %
Ventilator 5 5 5 2 5 3 2 3 3 3 1 377 93C-arm 3 3 4 2 5 3 3 3 3 2 2 316 78Monitor 3 3 3 3 3 2 3 3 2 1 1 269 66Centrifuge 2 3 3 3 2 2 1 3 1 1 1 228 56Scale 1 1 1 1 3 1 1 2 1 2 1 121 30
Fig. 5. PM priority index groups based on PS value of the proposed QFDmodel.
The complexity model [16] assesses the technical complexity 242
of a device considering four factors: equipment maintainability, 243
installation requirements, repair, and connectivity. The scores 244
are given for each factor range from 0 to 2 for evaluation based 245
upon the complexity level for each device in the list. 246
IV. RESULTS 247
For model verification, we utilized a dataset of 200 medical 248
equipment of two hospitals with one management system in 249
Piedmont province, Italy. Seventy different types of equipment 250
belonging to 17 different kinds of departments were analyzed 251
along data collected for one year (2012). Table III shows a 252
sample data of various types of investigated equipment along 253
with PS based upon our proposed model. 254
The scores in Table III are obtained referring to the scores that 255
are given in Table II. For instance, considering the “Ventilator” 256
device, and according to concept domain, “Function” is life sup- 257
port, “Physical Risk” is death, “Maintenance Requirements” is 258
extensive, “Area Criticality” is urgent, and “Device Criticality” 259
is critical; meanwhile, other parameters “Failure Rate,” “Useful 260
Life,” “Missed Maintenance,” and “Downtime Ratio” are de- 261
pending on the actual status of each device. Device complexity 262
is calculated relied on complexity model [16]. Accordingly, by 263
utilizing formula (2), the PS for every device is calculated as 264
shown in Table III. 265
By using the result of PS percentages, the PM priority is 266
classified into five categories as illustrated in Fig. 5. The first 267
class is very high-priority class and includes equipment that 268
supposed to be maintained within two weeks with PS percentage 269
equal or greater than 80. In second class, high-priority PM 270
should be performed within one month if priority percentage 271
in range 70–80. Class 3 is medium priority and contains all 272
IEEE
Proo
f
6 IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS, VOL. 00, NO. 00, 2014
Fig. 6. Results of the PM prioritization index for investigated medical equip-ment.
equipment that should be considered for PM within two months273
in case of priority percentage in range 60–70.274
Class 4 is low priority and includes all equipment with pri-275
ority percentage of 50 to 60, and in this case, PM should be276
performed within three months. Finally, all equipment with pri-277
ority percentage less than 50 could be visually inspected and278
considered for next PM as minimal PM.279
In our dataset of equipment and according to the proposed280
model and priority classification, the results indicate that 30281
devices (15%) need very high priority PM, 39 devices (19%)282
should be included as high priority, 59 devices (30%) should be283
considered for medium priority, 54 devices (27%) for low prior-284
ity, and finally, 18 devices (9%) should be with minimal priority285
PM. Fig. 6 presents dataset output classification considering the286
QFD model.287
By analyzing the results, very high priority class incor-288
porates all equipment with high-risk criteria, relatively high-289
mission-based criteria, and also with high-complexity level.290
High-priority class contains relatively high-risk criteria and291
mission-based criteria in addition to high missed maintenance.292
Medium priority class is considered for equipment with rela-293
tively high utilization level, area criticality, and old equipment.294
Low-priority class contains old not risky devices. Relatively sta-295
ble equipment does not need PM. The results are consistent with296
the classification given by an experienced clinical engineer. In297
other words, the devices that pose risk for patients and users in298
case of PM omission, old devices, and complex devices such as299
radiology equipment are highly considered for PM; meanwhile,300
low risk devices, reliable devices, and relatively new devices are301
modestly considered for PM. Moreover, as indicated in Fig. 5,302
the risk-based criteria contribute to approximately 45% of the303
weight of priority index. In fact, this contribution reflects high304
correlation existence between risk assessment and PM manage-305
ment of medical equipment.306
V. CONCLUSION307
In this study, QFD is presented for the first time to solve the308
problem of PM prioritization of medical equipment. The pro-309
posed model has proven its validity in real environment correctly310
separating equipment that needs PM from those that do not need311
it. The model was tested based upon the periodical schedule of312
the hospitals. It is important to note that the classification is 313
founded considering the requirements of patients and clinical 314
staff. By analyzing the results, we can state that the risk-based 315
criteria have a great impact on PM prioritization decision in ad- 316
dition to criticality and age of medical equipment. Also, the work 317
highlights the importance of existence of a detailed history for 318
every device that helps the decision makers to manage the med- 319
ical equipment obviously. The model can be used every time a 320
PM is to be planned modifying only the instruments data. More- 321
over, the model development could be improved by addressing 322
the customer requirements according to Kano’s model to clar- 323
ify the attitudes of customer satisfactions in medical equipment 324
PM. In addition, the developed QFD model can be implemented 325
in other stages of management of medical equipment such as 326
acquisition and procurement of medical equipment. The model 327
could be extended to maintenance agencies in order to organize 328
their PM programs. 329
REFERENCES 330
[1] W. A. Hyman, “The Theory and practice of preventive maintenance,” 331J. Clin. Eng., vol. 28, no. 1, pp. 31–36. Jan.–Mar. 2003. 332
[2] M. Ridgway, “Optimizing our PM programs,” Biomed. Instrum. Technol., 333vol. 43, no. 3, pp. 244–254. May/Jun. 2009. 334
[3] R. B. Arslan and Y. Ulgen, “Smart IPM: An adaptive tool for the preventive 335maintenance of medical equipment,” in Proc. IEEE 23rd Annu. Int. Conf. 336Eng. Med. Biol. Soc., Istanbul, Turkey, 2001, vol. 4, pp. 3950–3953. 337
[4] J. Ni and X. Jin, “Decision support systems for effective maintenance 338operations,” CIRP Ann.—Manuf. Technol., vol. 61, pp. 411–414, 2012. 339
[5] J. Joseph and S. Madhukumar, “A novel approach to data driven preventive 340maintenance scheduling of medical instruments,” in Proc. Int. Conf. Syst. 341Med. Biol., Kharagpur, India, 2010, pp. 193–197. 342
[6] R. Miniati, F. Dori, and G. B. Gentili, “Design of a decision support system 343for preventive maintenance planning in health structures,” Technol. Health 344Care J., vol. 20, pp. 205–214, 2012. 345
[7] S. Taghipour, D. Banjevic, and A. K. S. Jardine, “Prioritization of medical 346equipment for maintenance decisions,” J. Oper. Res. Soc., vol. 62, no. 9, 347pp. 1666–1687, 2010. 348
[8] S. O. Duffuaa, A. H. Al-Ghamdi, and A. Al-Amer, “Quality function 349deployment in maintenance work planning process,” in Proc. 6th Saudi 350Eng. Conf., 2002, vol. 4, pp. 503–512. 351
[9] B. M. Deros, N, Rahman, M. N. A. Rahman, A. R. Ismail, and 352A. H. Said, “Application of quality function deployment to study criti- 353cal service quality characteristics and performance measures,” Eur. J. Sci. 354Res., vol. 33, no. 3, pp. 398–410, 2009. 355
[10] S. Bennur and B. Jin, “A conceptual process of implementing quality 356apparel retail store attributes: An application of Kano’s model and the 357quality function deployment approach,” Int. J. Bus., Humanities Technol., 358vol. 2, no. 1, pp. 174–183, 2012 359
[11] X. X. Shen, K. C. Tan, and M. Xie, “An integrated approach to innovative 360product development using Kano’s model and QFD,” Eur. J. Innovative 361Manag., vol. 3, no. 2, pp. 91–99, 2000. 362
[12] D. J. Delgado, K. E. Bampton, and E. Aspinwall, “Quality function 363deployment in construction,” Constr. Manag. Econ. J., vol. 24, no. 6, 364pp. 597–609, 2007. 365
[13] L.K. Chan and M.-L. Wu, “A systematic approach to quality function 366deployment with a full illustrative example,” Int. J. Manag. Sci., vol. 33, 367no. 2, pp. 119–139, 2005. 368
[14] A. Al-Bashir, M. Al-Rawashdeh, R. Al-Hadithi, A. Al-Ghandoor, and M. 369Barghash, “Building medical devices maintenance system through quality 370function deployment,” Jordan J. Mech. Ind. Eng., vol. 25, no. 1, pp. 25–36, 371Feb. 2012. 372
[15] N. Z. Haron and F. L. M. Kairudin, “The application of quality function 373deployment (QFD) in the design phase of industrialized building system 374(IBS) apartment construction project,” Eur. Int. J. Sci. Technol., vol. 1, 375no. 3, pp. 56–66, 2012. 376
[16] N. F. Youssef and W. A. Hyman, “A medical device complexity model: A 377new approach to medical equipment management,” J. Clin. Eng., vol. 34, 378no. 2, pp. 94–98, 2009. 379
IEEE
Proo
f
SALEH et al.: PM PRIORITIZATION INDEX OF MEDICAL EQUIPMENT USING QFD 7
Neven Saleh received the B.Sc. degree in electronics and electrical commu-380nication and the diploma degree in electrical communication from Mansoura381University, Mansoura, Egypt, in 1999 and 2001, respectively, and the M.Sc.382degree in systems and biomedical engineering from Cairo University, Giza,383Egypt, in 2011. She is currently working toward the Ph.D. degree in Biomedical384Engineering from Cairo University, and also from the Politecnico di Torino,385Turin, Italy, since she was granted a channel scholarship.386
Since 2001, she has been a Clinical Engineer with Mansoura University387Children Hospital, Mansoura.388
389
Amr A. Sharawi received the B.Sc. degree in electronics and communication390and the M.Sc. and Ph.D. degrees in systems and biomedical engineering, all391from Cairo University, Giza, Egypt, in 1976, 1981, and 1991, respectively.392
Since 2001, he has been an Associate Professor in biomedical engineering,393Faculty of Engineering, Cairo University.394
395
Manal Abd Elwahed received the B.Sc., M.Sc., and Ph.D. degrees in biomedi-396cal engineering from the Faculty of Engineering, Cairo University, Giza, Egypt,397in 1987, 1992, and1999, respectively.398
She is currently an Associate Professor in the Systems and Biomedical En-399gineering Department, Faculty of Engineering, Cairo University.400
401
Alberto Petti received the first M.Sc. degree in biomedical engineering and 402the second M.Sc. degree in management and health technologies from the 403Politecnico di Torrino, Torino, Italy, in 2007 and 2008, respectively. 404
Since 2000, he has been responsible for the clinical engineering service in 405the Public Hospital of Biella, Italy. He is currently involved in realization of the 406new hospital of the city. He is currently a Mechanical Engineer. His research 407interests include the applications of methods for better organizations of the 408public health system, involving technologies and ICT tools, particularly related 409to management of health technologies. 410
411
Daniele Puppato received the B.Sc. and M.Sc. degrees in biomedical engineer- 412ing from the Politecnico di Torrino, Torino, Italy, in 2003 and 2005, respectively. 413
From 2006 to June 2013, he was with AReSS Piemonte as a Health Care 414Technology Management Leader. He is currently working as a Consultant in a 415leading Italian company developing software solutions for technical manage- 416ment of hospitals. 417
Mr. Puppato is a Member of the Italian Association of Clinical Engineers. 418419
Gabriella Balestra received the M.Sc. degree in computer science from the 420University of Turin, Turin, Italy, in 1980, and the Ph.D. degree in computer and 421system engineering from the Politecnico di Torino, Turin, in 1989. 422
She is currently an Assistant Professor in the Department of Electronics 423and Telecommunications, Politecnico di Torino, where she is a Lecturer for 424the Biomedical Engineering course. She also teaches decision support systems, 425biomedical data classification techniques, and medical informatics. 426
Dr. Balestra is a member of GNB, the Italian National Group of Bioengi- 427neering, and the Association for the Advancement of Medical Instrumentation. Q2428
429
IEEE
Proo
f
QUERIES430
Q1. Author: Fig. 2 is not cited in the text. Please cite it at an appropriate place.431
Q2. Author: Please provide the expansion of “GNB.”432
