Using a Hybrid Simulation Model to Maximize Patient Throughput

of Magnetic Resonance Imaging in a Medical Center

Li-Hui Lee

1 a

, Yang-Yu Ou

, Yu-Ting Cheng

, Ying-Chou Sun

, Hsiu-Mei Wu

and

Wan-You Guo

Department of Health Care Management, National Taipei University of Nursing and Health Sciences, No. 89,

Neijiang Street, Wanhua District, Taipei City, Taiwan

Department of Radiology, Taipei Veterans General Hospital, No.201, Sec. 2, Shipai Rd., Beitou Dist.,

Taipei City 112, Taiwan

Keywords: Agent-based, Magnetic Resonance Imaging, Waiting Time, Throughput, Cost-effectiveness Analysis.

Abstract: The purpose of this study is to shorten the waiting time of a patient to undergo a magnetic resonance imaging

(MRI) examination as well as to increase the utilization rates of the five MRI scanners in a 2,735-bed medical

center of Taiwan. We simulate the current MRI examination workflow through a hybrid simulation model,

consisting of discrete event and agent-based simulations. Then, we maximize the MRI patient throughput per

day with more radiographers during current business hours. For reasons of feasibility, a cost-effectiveness

analysis is also conducted. AnyLogic, Excel, and SAS were used for building the simulation models and

conducting the statistical analysis. The results show that providing 248 additional patient examinations with

one additional radiographer employed during the 11:00 to 19:00 time frame would shorten the waiting time

of a patient to undergo an MRI examination by 2.51 days, increase the utilization rate of each MRI scanner

by an average of 6.15%, and bring an additional gross income of US$38,424 to US$69,169 per 31-day month.

1 INTRODUCTION

For decades, high-tech medical imaging scanners,

such as magnetic resonance imaging (MRI), have

been widely used in the examination, diagnosis and

tracking of various diseases and in the detection of

cancer stages and the determination of cancer

metastasis position. With the improvement of

personal health awareness and the rapid

dissemination of health-related information through

the Internet, people pay more attention to their health

status and that of their family members than before.

Precise preventive photographic examination in

medical institutions or health examination centers is

becoming increasingly common (Global Views, 2015,

National Health Insurance Administration Ministry

of Health and Welfare, 2018b).

To meet the needs of MRI examination, the

number of medical imaging scanners in some

countries has been increasing yearly. According to

the Organization for Economic Co-operation and

Development (OECD) statistics up to 2017, the U.S.

https://orcid.org/0000-0003-0377-7459

has more scanners than any other European or

American country, at 37.6 scanners per million

people. Germany and Italy are in second and third

places at 34.5 and 28.4 scanners per million people,

respectively. Asia ranks first in the world with 51.7

scanners per million people in Japan, followed by

29.1 scanners per million people in South Korea.

According to the statistics of the Central Health

Insurance Agency of the Ministry of Health and

Welfare, the number of scanners in Taiwan had

increased to 226 by the second half of 2017, with

approximately 9.58 scanners per million people,

which is between Slovakia (9 scanners per million

people) and Canada (10 scanners per million people)

(National Health Insurance Administration Ministry

of Health and Welfare, 2018a, The Organisation for

Economic Co-operation and Development, 2017).

Although the number of MRI scanners has

increased yearly, the waiting time for the examination

is still too long. According to the standards in Ontario,

Canada, patients with preferential treatment

classifications P1 and P2 are very urgent outpatients,

inpatients, and Emergency Department (ED) patients

Lee, L., Ou, Y., Cheng, Y., Sun, Y., Wu, H. and Guo, W.

Using a Hybrid Simulation Model to Maximize Patient Throughput of Magnetic Resonance Imaging in a Medical Center.

DOI: 10.5220/0007932003090316

In Proceedings of the 9th International Conference on Simulation and Modeling Methodologies, Technologies and Applications (SIMULTECH 2019), pages 309-316

ISBN: 978-989-758-381-0

309

who should be examined immediately or within 48

hours; the classification P3 denotes emergency

outpatients who should be examined in 2 to 10 days;

and the classification of P4 denotes general

outpatients. The waiting time for MRI examinations

should not exceed 28 days. However, the average

MRI waiting time for adult outpatients is 59 days

(Health Quality Ontario, 2018). The waiting time in

Manitoba, Canada, is approximately 68 workdays

(Province of Manitoba, 2018). Taiwan currently has

no similar statistics for the waiting time for MRI

examination. However, the waiting time for MRI

examination is still recognized as long.

In the health care management field, to understand

the effects of new policies and new technology

introductions, organizations may need to analyze

relevant economic roles (e.g., competitors and

suppliers), responses from users, and environmental

impact. However, exploring the effectiveness of a

new strategy by traditional techniques has its

limitations. Most of the traditional techniques can

only analyze data theoretically or qualitatively. Even

some new strategies require actual or trial investment

in human and material resources so that managers can

understand the impact and determine whether a

strategy deserves continued investment. To avoid the

losses caused by failure after investment, many

simulation analyses have been applied to ED research

since 1999 (Hurwitz et al., 2014, Saoud et al., 2016).

Additionally, a small number of studies have been

applied to rehabilitation (New et al., 2015),

orthopedics (Rohleder et al., 2011), surgery (Sobolev

et al., 2011), hospitalization (Hahn-Goldberg et al.,

2014), ophthalmology and radiology (Lindsköld et

al., 2012, Viana, 2014). All these studies show that a

good simulation model is adaptable (Paranjape, 2009)

and can be adapted to practice changes as an aid to

the evaluation decision before the new strategy is

adopted.

A patient’s waiting time for an examination in a

radiology department includes the duration from

when a radiology request is made in the clinic, the

radiology department receives the request, the

radiology department vets the request, the radiology

department schedules the examination, to when the

patient attends, waits and completes the examination

(Olisemeke et al., 2014). Retrospective to 1987, a

study simulated the daily non-admission patients

through the radiology department of a large acute care

hospital. This study showed that the addition of one

more radiologist would lead to a reduction in the

length of stay of non-admission patients (Klafehn,

1987).

Later studies focused on radiology department

services, such as mammography (Coelli et al., 2007),

sonography (Johnston et al., 2009), computed

tomography (CT) (Ramakrishnan et al., 2004, van

Lent et al., 2012), and X-ray (Oh et al., 2011,

Lindsköld et al., 2012),. The examination process has

been simulated to explore the relevant key

performance indicators (KPIs) before and after the

improvement plan. These studies assessed the number

of patients examined within one hour, the time of

completing image reports, the time of the patient

waits for the examination, the length of time the

patient stays in the department, and the utilization rate

of the radiologist as effective indicators as to assist in

the formulation and implementation of preplanning

decisions (Ramakrishnan et al., 2004, Coelli et Al.,

2007, Johnston et al., 2009, Oh et al., 2011). The MRI

examination workflow and KPIs are different in

different organizations and units. Additionally, an

MRI scanner cannot provide all examination services.

It depends on if the needed coil types are adopted by

the scanner or not. However, a review of past studies

in radiology, mostly using discrete event simulation

as the main analytical method, the characteristic of

patient, staff, and scanner were lack to control and

define in a simulated workflow. The KPIs of the

individual patient, scanner and the department could

not be accurately estimated.

Hence, the purpose of the study was to propose a

new strategy to maximize patients throughput so that

the waiting time of a patient to undergo a magnetic

MRI examination can be shortened as well as the

utilization rates of MRI scanners can be increased.

The specific aims in this study are (a) to develop a

discrete event (DES) and agent-based simulations

(ABS) model to simulate the MRI examination

workflow at an MRI department in a medical center

of Taiwan, and (b) to identify which time frames of

day have fewer patients examined, to experiment with

the proposed strategy by hiring radiographers in those

time periods in the simulation model, and then to

explore the changes in KPIs, including the daily MRI

scanner utilization rate, monthly gross income, and

waiting time to undergo an MRI examination.

2 RADIOLOGY DEPARTMENT

SETTING

2.1 MRI Examination Services

The research setting was the MRI department, under

the radiology department, of Taipei Veterans General

Hospital (TVGH), which is a 2,735-bed medical

center in northern Taiwan. This department provides

MRI examination services for ED patients, inpatients,

and outpatients. The MRI examination service runs

SIMULTECH 2019 - 9th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

310

Figure 1: The patient flow of MRI examination.

from 7:30 to 23:00 on workdays and from 8:00 to 23:00

on weekends. The shifts are different for four types of

human resources, including administrators, nurses,

radiographers, and medical resident. The administrator

and nurse only work the day shift from 8:00 to 17:30.

Nevertheless, radiographers need work in either the

day shift from 7:00 to 15:00 or the night shift from

15:00 to 23:00. Medical residents work shifts

depending on the departmental regulations, but they

are available to support the MRI examination services

from 7:00 or 8:00 to 23:00, depending on the day.

The duties of each worker are as follows:

 An administrator is responsible for examination

date and time scheduling, registration affairs and

explanation of related matters for patients.

 Nurses are responsible for intravenous (IV)

injection and assisting the administrator.

 Radiographers are responsible for MRI scanner

operation and MRI examination execution.

 Medical residents are responsible for

radiocontrast agent injection and the tasks of

nurses when nurses are off work, i.e., before 8:00

and after 17:30.

The fixed resources of the MRI department

include one 3.0 Tesla (3.0-T) MRI scanner (Scanner

A) and four 1.5 Tesla (1.5-T) MRI scanners (Scanner

B to E). The former is for a specific examination.

Patients will be assigned to the scanner after check-in

if they schedule the 3.0-T MRI service. The latter is

for general examinations. Patients will be randomly

assigned to one of four available scanners after check-

in if they schedule for a 1.5-T MRI examination.

In the MRI department, an average of 91 patients

(92 and 89 patients on weekdays and weekends,

respectively) are examined a day, i.e., 18 patients for

each scanner. However, the need for MRI

examination service exceeds the supply. When an ED

patient, inpatient, or outpatient needs an MRI

examination service, he or she must wait to be

examined for five hours, three days, or 30 days,

respectively. The utilization rate of our five scanners

ranges from 55% to 73%. Therefore, how to reduce

the idle condition of the scanner and to increase the

number of served patients each day warrants attention.

2.2 MRI Examination Flow

Figure 1 is the patient flow of MRI examination in the

TVGH. First, patients have to check in at the

registration counter and sign the consent form for MRI

examinations. Then, patients walk to the dressing room

to change into loose-fitting hospital clothes. If patients

need a radiocontrast agent injected, the nurse will then

perform the set up for intravenous placement. All

patients will wait in the waiting area until the

radiographer calls their names to enter the MRI

examination room. Then, the patients will be examined

without radiocontrast agent injection or examined

before and after radiocontrast agent injection by

medical residents. The examined patients with an IV

set will have the IV set removed. All patients will

change back into their clothes before leaving.

3 MODEL DEVELOPMENT AND

EVALUATION

3.1 Simulation Framework

The simulation framework applies DES and ABS

(Hamrock et al., 2013, Viana et al., 2018). The DES is

used for MRI examination flow, the MRI scanners, and

staffs. The ABS is applied when the patient with the

Using a Hybrid Simulation Model to Maximize Patient Throughput of Magnetic Resonance Imaging in a Medical Center

311

different need of MRI coil types interact with each

other by competing for MRI scanners. The patients

have been modeled as agents to allow future

development and provide a base for future expansion

of the model.

As shown in Figure 2, there are two types of input

parameters. One parameter is the patient attributes, and

the other parameter is MRI examinations-related

demand, services and resources. The former includes

patients’ gender, the utilized medical service types (ED

patients, inpatient, or outpatients), need for a

radiocontrast agent injection or not, utilized MRI

scanner type (3.0-T or 1.5-T), and utilized MRI coil

type (11 types). The latter includes an hourly number

of patient arrivals, daily MRI examination duration

time in minutes (DailyMRIexamTime), human

resources (one administrator, one nurse, and ten

radiographers), and one 3.0-T and four 1.5-T MRI

scanners. For DailyMRIexamTime, if a patient needs

radiocontrast agent injection, DailyMRIexamTime

will encompass the injection of the radiocontrast agent

as well as the coil and scanner preparation and the MRI

examination execution. Otherwise, the

DailyMRIexamTime includes only time in the coil and

scanner preparation, and examination execution.

Output parameters are composed of four KPIs,

which are the average daily utilization rates of each

MRI scanner (DailyUtiRate, i.e., each MRI scanner’s

examination duration time per day divided by 16

hours), average daily overtime minutes

(DailyOverTime), the monthly gross income

(MonthlyGrossIncome, i.e., the increased revenue

due to the increased number of patients minus the cost

of the increased number of radiographers per month),

and the shortened waiting time (ShortenWaitDays)

for a patient who originally would have to wait for 30

days, i.e., an outpatient’ waiting days for an MRI

examination in the TVGH, to be underwent an MRI

examination.

The study first built a baseline model to simulate

the current MRI examination workflow in the MRI

department in the TVGH. To validate if the model with

simulated data is the same as the model with raw data,

the Wilcoxon rank-sum test was used for comparing

three input parameters and two output KPIs. The

former includes a daily number of patient arrivals

(DailyNumPatients), the number of patients utilizing

the three medical service types

(DailyNumServiceTypes) and DailyMRIexamTime.

The later includes DailyUtiRate and DailyOverTime.

After confirmation that the baseline model with

simulated data has a statistically nonsignificant

difference from the baseline model with raw data, we

conduct experiments to build the proposed model

based on the baseline model with simulated data. The

proposed model is developed by adding numbers of

patients and staffs per day to find an optimal

DailyUtiRate with statistically nonsignificant

DailyOverTime. Additionally, we not only validate

that DailyOverTime has a statistically nonsignificant

difference from the baseline model with simulated data,

but also confirm that DailyUtiRate,

MonthlyGrossIncome, and ShortenWaitDays are

positive results in the proposed model.

3.2 Material and Data Preprocessing

The 2,814 MRI examination logs with de-

identification of patients and radiographers in

December of 2016 are collected from the MRI

department. One log that belonged to the MRI

examination in November is excluded. All data fields

include patients’ gender, utilized medical service

types, need for a radiocontrast agent injection or not,

utilized MRI scanner (labeled A to E), the received

MRI examination name (96 classifications), and

Figure 2: Simulation framework of MRI examinations workflow.

SIMULTECH 2019 - 9th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

312

DateTime of getting on and off the MRI examination

table. Other information, such as workflow, business

hours, human resources allocation, the number and

location of 3.0-T and 1.5-T scanners are informed by

authors Sun, Wu, and Guo according to their work

experience. This study was approved by the Taipei

Veterans General Hospital Institutional Review

Board in December 2018 (2018-01-010CC).

To generate input parameters and output KPIs,

data preprocessing was conducted. To produce an

hourly number of patient arrivals on each day, the

study presumed that arrival time was 40 minutes

before the DateTime of patients getting on the MRI

examination table. Additionally, the utilized MRI

scanners are converted into two types of MRI scanner

types (scanner A is 3.0-T, while scanners B to E are

1.5-T). Distinct 96 MRI examination names are

converted into 11 coil types. DailyMRIexamTime

equals to the DateTime of getting off minus getting

on the MRI examination table. DailyOverTime equals

to the DateTime of getting off the MRI examination

table minus 23:00.

3.3 Input Parameters

Regarding the patient attributes in the simulation

models, 49.6% of patients were male, according to the

raw data. The percentage of ED patients, inpatients,

and outpatients are 5.6%, 19.1%, and 75.3%,

respectively. Some 57.96% of all patients need

radiocontrast agent injection for the MRI examination.

The MRI scanner A (3.0-T) is pre-arranged for 13%

of all patients, while MRI scanners B to E (1.5-T) are

pre-arranged for 87% of all patients. The eleven types

of MRI examination coil have different utilization

rate, i.e., standard head (1.07%), torso (16.71%),

shoulder (1.78%), 8-inch head (51.21%), low

extremity (0.07%), neck (6.83%), torso head flexible

(2.67%), breast (0.71%), spine (14.05%), cardiac

(1.03%), and knee (3.88%).

The MRI examination service applies the first in,

first out (FIFO) method. For each patient to be

examined, the MRI scanner B, C, D, or E is arranged

according to what coil type the patient needs. Each

scanner’s monthly examination service volume in

different coil types from high to low decides the

scanner’s priority for the patient. For example, if the

patient needs coil type “cardiac” and scanners’

service volume for using the cardiac coil from high to

low are scanners D, B, C, and E, the patient will be

arranged to the available MRI scanner D, B, C, or E,

accordingly.

For the MRI examinations-related demand, an

hourly number of patient arrivals (Table 1) is

calculated using the MRI examination records of

December 2016. The DailyMRIexamTime for each

scanner is used to generate a custom distribution.

Additionally, the study adopts the thumb rule by

using the triangular distribution (setting minimum,

maximum and mode in minutes) for configurations of

the time registration service (1,3,5), clothes changing

1,3,2), IV setting up (1,4,3), and IV set removing

(1,3,2).

3.4 Output Key Performance

Indicators (KPIs)

As mentioned above, MonthlyGrossIncome is

estimated by the increased monthly revenue minus

increased monthly cost of the proposed solution. The

increased monthly revenue equals the increased

number of patients multiplied by the points of the

hospital global budget payment (pts) for providing

each MRI examination to a patient without

radiocontrast (6,500 pts) or with radiocontrast agent

injection (11,500 pts) (Ministry of Health and

Welfare, 2017). The monetary value of points was

converted at a 0.8265 ratio and a rate of 30 Taiwanese

dollars to one US dollar (National Health Insurance

Committee Ministry of Health and Welfare, 2018).

The increased monthly cost equals the increased

number of radiographers multiplied by US$1,546,

which is a market salary for a radiographer with a

postgraduate degree.

3.5 The Proposed MRI Examination

Model

Based on the baseline model, the study experiments

with seven configurations by gradually adding even

numbers of patients (i.e., 2, 4, 6, up to 14 patients) a

day to the proposed model. We chose even numbers

of patients because fewer patients were examined at

two-time slots (i.e., 11:00, during lunch, and 17:00,

during dinner) each day. To satisfy the examination

needs of patients, the proposed model is designed by

hiring one additional radiographer from 11:00 to

19:00 per month to provide service to these added

patients.

Table 1: Number of patient arrivals per hour.

Time

6:00

7:00

8:00

9:00

10:00

11:00

12:00

13:00

14:00

15:00

16:00

17:00

18:00

19:00

20:00

21:00

22:00

Number

of patients

Using a Hybrid Simulation Model to Maximize Patient Throughput of Magnetic Resonance Imaging in a Medical Center

313

3.6 The Proposed Model Validation

and Analysis Tool

The simulated dates for the baseline and proposed

models are from 1 December 2016 to 1 January 2017.

According to the simulation framework (in section

3.1), the study tests the baseline model and the

proposed model with simulated data to confirm that

both have nonsignificant differences in

DailyOverTime. The statistical test method used is

the Wilcoxon rank sum test. The study used flow

charts for workflow analysis, AnyLogic for

simulation model development, Excel for exported

logs of AnyLogic, and SAS for statistical analysis.

4 SIMULATION RESULTS

As shown in Table 2, the baseline model with raw

data and simulated data have nonsignificant

differences in DailyNumPatients,

DailyNumServiceTypes, DailyMRIexamTime,

DailyUtiRate, and DailyOverTime. In the raw and

simulated data, DailyNumPatients are 2,813 and

2,821 patients, respectively. DailyNumServiceType,

including ED patient, inpatient, and outpatient, are

5.62, 19.09, and 75.29%, respectively, in the raw data

and 5.28, 19.00, and 75.72%, respectively, in the

simulated data DailyMRIexamTime is 35.78 and

36.10 minutes, respectively.

Table 2: The validation of the baseline model.

Statistical parameters

Baseline

model with

raw data

Baseline

model with

simulated

data

p-value

DailyNumPatients: n

2,813

2,821

0.6384

DailyNumServiceTypes: n (%)

ED patients

158(05.62)

149(05.28)

0.7916

Inpatients

537(19.09)

536(19.00)

0.8433

Outpatients

2,118 (75.29)

2,136(75.72)

0.8820

DailyMRIexamTime: mean in minutes (standard

deviation)

35.78(1.98)

36.10(1.38)

0.5685

DailyUtiRate: (%)

MRI scanner A

55.10

50.41

0.1570

MRI scanner B

74.20

76.19

0.7514

MRI scanner C

74.39

75.50

0.3638

MRI scanner D

74.85

74.52

0.2050

MRI scanner E

73.35

76.22

0.3107

DailyOverTime: median in minutes (standard deviation)

0(42.13)

0(10.35)

0.9300

DailyUtiRate for each scanner from A to E is 55.10,

74.20, 74.39, 74.85, and 73.35%, respectively, in the

raw data and 50.41, 76.19, 75.50, 74.52, and 76.22%,

respectively, in the simulated data. DailyOverTime is

0 in both models.

As shown in Table 3, when the number of daily

patients is increased to greater than or equal to eight

patients, at least one scanner’s DailyUtiRate can

Table 3: One-month simulation in the proposed model.

Note:

†

without contrast: number of monthly patients multiplied by points of the hospital global budget payment (pts) for

providing each MRI examination to a patient without radiocontrast agent injection (6,500 pts);

‡

with contrast: number of

monthly patients multiplied by points of the hospital global budget payment (pts) for providing each MRI examination to a

patient with radiocontrast agent injection (11,500 pts).

Experi-

ments

:Added

No. of

patient

arrivals

per day

Total

No. of

patients

in the

month

Utilization rate

of the MRI scanner

(daily) (%, used

hours/16 hours)

Average

daily

overtime

minutes

Added

No. of

radio-

grapher

per

month

①

Increased

staff cost

(US$)

per

month

②

Increased point of

hospital global

budget payment

(pts) per month

(31 days)

③

Increased

gross income per

month (31 days)

【② x 0.8265

x 0.03US-①】

Shortened

days of

patient

waiting to

schedule an

MRI

examination

(1 working

day=16

hours)

without

contrast

†

with

contrast

‡

without

contrast

with

contrast

Baseline

2,821

2,883

1,546

403,000

713,000

8,446

16,133

11H

2,945

1,546

806,000

1,426,000

18,439

33,812

1D & 5H

3,007

1,546

1,209,000

2,139,000

28,431

51,491

1D & 15H

3,069

1,546

1,612,000

2,852,000

38,424

69,169

2D & 8H

+10

3,131

1,546

2,015,000

3,565,000

48,416

86,848

3D & 1H

+12

3,193

1,546

2,418,000

4,278,000

58,408

104,527

3D & 10H

+14

3,255

1,546

2,821,000

4,991,000

68,401

122,206

4D & 2H

SIMULTECH 2019 - 9th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

314

reach more than 83%. However, only when the

proposed model adding eight patients a day has a

statistically nonsignificant difference in

DailyOverTime (median=0 minute) compared to the

raw data (p=.3257). The other experiments increase

DailyOverTime.

To hire one additional radiographer will increase

the cost by US$1,546 per month. The increased

monthly hospital global budget payment points

(income) due to MRI examination without and with

radiocontrast agent injection are 1,612,000 and

2,852,000 points, respectively. As a result,

MonthlyGrossIncome will be between US$38,242

and US$69,169. For a patient who originally has to

wait for 30 days to be arranged an MRI examination,

under this situation, the patient can be examined two

days and eight hours sooner.

5 DISCUSSION AND

CONCLUSION

Here, we report preliminary research on the

maximization of MRI examination patient throughput.

A major finding is that to hire one additional

radiographer for MRI examination in the TVGH

would increase the serviced volume by 248 patients a

month (31 days). This proposed solution would

increase the utilization rate of scanners, raise the

monthly gross income, and shorten the waiting time

for patients who need the MRI examination service.

Except for scanner A, the other scanners have a

relatively high utilization rate. This observation is

because scanner A is also used for research purposes.

Compared to the discrete event-based simulation, the

hybrid method can not only simulate the MRI

examination workflow but also include

characteristics of patients, staffs, and scanners. This

characteristic makes the simulation model easy to

adjust to service logic according to the agent

characteristics, e.g., the utilized coil type, and lead to

a nonsignificant difference from the reality.

This is also the first study in Taiwan to use

simulation analysis methods to analyze and improve

the evaluation of the current situation of the MRI

department. In addition to proposing different

opinions on improving the evaluation of the program

from the traditional improvement methods, this

simulation also assists the managers of the MRI

department in evaluating the feasibility of the

improvement program in practice and provides

different decision-making aids for managers as a

reference.

Despite the hybrid simulation model’s

advantages, it does have four limitations. First, the

time stamps of patient registration, clothes changing,

IV setting up and IV set removal cannot be obtained.

We calculate these timestamps according to the

thumb rule. Moreover, the DateTime of getting on

and off the MRI examination table were manually

entered into an information system by radiographers.

The accuracy of our data is unknown, although the

simulated model is well developed. Third, the

simulation model is based on the data during

December 2016. The variances due to seasonal

reasons are not accounted for in the study. Fourth, the

simulated model cannot be generalized to other fields.

However, the study design and simulation method,

including the parameters, can be a reference for

related workflows, e.g., CT examinations.

This maximization of MRI examination patient

throughput is still in the experimental stage, and much

more work has yet to be conducted. Much more also

needs to be known about how to pre-arrange patients

with different coil-type needs and how to properly

assign on-site patients to busy scanners. This study

should provide a simulation basis for additional

research. There is a continuing need for an adequate

solution with big-data simulations for the practical

application of MRI examination management with a

high utilization rate of scanners and less waiting time

for undergoing an examination.

ACKNOWLEDGMENTS

This study is supported by the Taiwan Ministry of

Science and Technology under grant number MOST

107-2221-E-227-002-.

REFERENCES

Coelli, F. C., Ferreira, R. B., Almeida, R. M. V. & Pereira,

W. C. A. 2007. Computer simulation and discrete-event

models in the analysis of a mammography clinic patient

flow. Computer methods and programs in biomedicine,

87, 201-207.

Global Views 2015. The investigation of physical

examination center in Taiwan.

Hahn-Goldberg, S., Chow, E., Appel, E., Ko, F., Tan, P.,

Gavin, M., Ng, T., Abrams, H., Casaubon, L. & Carter,

M. 2014. Discrete event simulation of patient

admissions to a neurovascular unit. Journal of

healthcare engineering, 5, 347-360.

Hamrock, E., Paige, K., Parks, J., Scheulen, J. & Levin, S.

2013. Discrete event simulation for healthcare

organizations: a tool for decision making. Journal of

Healthcare Management, 58, 110-124.

Health Quality Ontario. 2018. WAIT TIMES FOR

DIAGNOSTIC IMAGING [Online]. Available:

https://www.hqontario.ca/System-Performance/Wait-

Using a Hybrid Simulation Model to Maximize Patient Throughput of Magnetic Resonance Imaging in a Medical Center

315

Times-for-Diagnostic-Imaging?ScanType=AdultMRI

[Accessed].

Hurwitz, J. E., Lee, J. A., Lopiano, K. K., Mckinley, S. A.,

Keesling, J. & Tyndall, J. A. 2014. A flexible

simulation platform to quantify and manage emergency

department crowding. BMC medical informatics and

decision making, 14, 50.

Johnston, M., Samaranayake, P., Dadich, A. & Fitzgerald,

J. Modelling radiology department operation using

discrete event simulation. MODSIM, International

Congress on Modelling and Simulation, Cairns, 2009.

678-84.

Klafehn, K. A. Impact points in patient flows through a

radiology department provided through simulation.

Proceedings of the 19th conference on Winter

simulation, 1987. ACM, 914-918.

Lindsköld, L., Wintell, M., Aspelin, P. & Lundberg, N.

2012. Simulation of radiology workflow and

throughput. Radiology management, 34.

Ministry of Health and Welfare. 2017. What is the meaning

of points in Annual Report? [Online]. Available:

https://dep.mohw.gov.tw/DOS/cp-1716-14931-

113.html [Accessed].

National Health Insurance Administration Ministry of

Health and Welfare. 2018a. The amount, brand, type,

date of manufacture, date of purchase and service life

of MRI sanner in Taiwan. [Online]. Available:

https://data.gov.tw/dataset/22238 [Accessed].

National Health Insurance Administration Ministry of

Health and Welfare. 2018b. The physical examination

popularizing rate of age 40-64 adult in Taiwan.

[Online]. Available: https://data.gov.tw/dataset/21274

[Accessed].

National Health Insurance Committee Ministry of Health

and Welfare. 2018. The total payment of national health

insurance medical consultation fee reference index

[Online]. Available: https://dep.mohw.gov.tw/NHIC/

lp-1665-116.html [Accessed].

New, P. W., Racp, F., Stockman, K., Cameron, P. A., Olver,

J. H. & Stoelwinder, J. U. 2015. Computer simulation

of improvements in hospital length of stay for

rehabilitation patients. Journal of rehabilitation

medicine, 47, 403-411.

Oh, H. C., Toh, H. G. & Giap Cheong, E. S. 2011.

Realization of process improvement at a diagnostic

radiology department with aid of simulation modeling.

Journal for Healthcare Quality, 33, 40-47.

Olisemeke, B., Chen, Y., Hemming, K. & Girling, A. 2014.

The effectiveness of service delivery initiatives at

improving patients’ waiting times in clinical radiology

departments: a systematic review. Journal of digital

imaging, 27, 751-778.

Paranjape, R. 2009. Multi-Agent Systems for Healthcare

Simulation and Modeling: Applications for System

Improvement: Applications for System Improvement,

IGI Global.

Province of Manitoba 2018. Health Services Wait Time

Information.

Ramakrishnan, S., Nagarkar, K., Degennaro, M., Srihari,

K., Courtney, A. K. & Emick, F. A study of the CT scan

area of a healthcare provider. Proceedings of the 36th

conference on Winter simulation, 2004. Winter

Simulation Conference, 2025-2031.

Rohleder, T. R., Lewkonia, P., Bischak, D. P., Duffy, P. &

Hendijani, R. 2011. Using simulation modeling to

improve patient flow at an outpatient orthopedic clinic.

Health care management science, 14, 135-145.

Saoud, M. S., Boubetra, A. & Attia, S. 2016. A Simulation

Knowledge Extraction-based Decision Support System

for the Healthcare Emergency Department.

International Journal of Healthcare Information

Systems and Informatics (IJHISI), 11, 19-37.

Sobolev, B. G., Sanchez, V. & Vasilakis, C. 2011.

Systematic review of the use of computer simulation

modeling of patient flow in surgical care. Journal of

medical systems, 35, 1-16.

The Organisation for Economic Co-Operation and

Development. 2017. Magnetic resonance imaging

(MRI) units [Online]. Available:

https://data.oecd.org/healtheqt/magnetic-resonance-

imaging-mri-units.htm [Accessed].

Van Lent, W. A., Deetman, J. W., Teertstra, H. J., Muller,

S. H., Hans, E. W. & Van Harten, W. H. 2012.

Reducing the throughput time of the diagnostic track

involving CT scanning with computer simulation.

European journal of radiology, 81, 3131-3140.

Viana, J. Reflections on two approaches to hybrid

simulation in healthcare. Simulation Conference

(WSC), 2014 Winter, 2014. IEEE, 1585-1596.

Viana, J., Simonsen, T. B., Dahl, F. A., & Flo, K. A hybrid

discrete event agent based overdue preganancy

outpatient clinical simulation model. Simulation

Conference (WSC), 20118 Winter, 2018. IEEE, 1488-

1499.

SIMULTECH 2019 - 9th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

316