Professional Documents
Culture Documents
The Application of Collaborative Business Intellig
The Application of Collaborative Business Intellig
Abstract
Background: We aimed to apply collaborative business intelligence (BI) system to hospital supply, processing and distribu-
tion (SPD) logistics management model.
Methods: We searched Engineering Village database, China National Knowledge Infrastructure (CNKI) and Google for
articles (Published from 2011 to 2016), books, Web pages, etc., to understand SPD and BI related theories and recent re-
search status. For the application of collaborative BI technology in the hospital SPD logistics management model, we rea-
lized this by leveraging data mining techniques to discover knowledge from complex data and collaborative techniques to
improve the theories of business process.
Results: For the application of BI system, we: (i) proposed a layered structure of collaborative BI system for intelligent
management in hospital logistics; (ii) built data warehouse for the collaborative BI system; (iii) improved data mining tech-
niques such as supporting vector machines (SVM) and swarm intelligence firefly algorithm to solve key problems in hospital
logistics collaborative BI system; (iv) researched the collaborative techniques oriented to data and business process optimiza-
tion to improve the business processes of hospital logistics management.
Conclusion: Proper combination of SPD model and BI system will improve the management of logistics in the hospitals.
The successful implementation of the study requires: (i) to innovate and improve the traditional SPD model and make ap-
propriate implement plans and schedules for the application of BI system according to the actual situations of hospitals; (ii)
the collaborative participation of internal departments in hospital including the department of information, logistics, nursing,
medical and financial; (iii) timely response of external suppliers.
Introduction
Hospital logistics is one of the application fields Their primarily focus was on the development of
for logistics. It mainly covers the procurement, management models in the field of procurement,
storage, distribution, usage and the control of inventory and distribution.
medical equipment, drugs and supplies in hospit- In recent years, several models have been pro-
al. Since 1950s, scholars have paid attention to posed by scholars in terms of procurement. La-
hospital material management, especially in the pierre et al. (4) designed procurement decision-
inventory management of medical materials. With making method of hospital materials with the
the development of supply chain and logistics tabu search algorithm. By analyzing the colla-
related theories, the hospital logistics attracted borative supply model between hospitals and
extensive attention, and many scholars started to medical suppliers, Centobelli et al. (5) proposed a
systematically explore the hospital logistics (1-3). more streamlined procurement strategy of medi-
cal supplies based on e-business. With the help of Business intelligence (BI) transfers various data
fuzzy failure mode and effect analysis (FMEA), from enterprises into information or knowledge
Kumru et al. (6) improved the procurement pro- displayed in a cooperative manner that enterprise
cedures and methods. When it comes to the in- managers interested in. They were able to provide
ventory management, Bijvank et al. (7) tried to scientific evidence for the enterprise managers'
optimize the hospital inventory by designing and decision-making and finally strengthen the com-
using the capacity model and ability model. Shan petitive advantage of enterprises. In the early
et al. (8) employed the method of greedy algo- stage, scholars primarily focused on the data
rithm to deal with the multi-level inventory prob- warehouse and data marts. Ariyachandra et al.
lems in the hospital. Other researchers (9-11), (16) put forward the architecture of data mart bus
proposed selecting the optimal distribution mod- and selected the data mart structure of consistent
el with the methods of process modeling and correlation dimension. One year later, they also
mixed linear programming modeling based on proposed an independent and separated architec-
the graph theory. ture for the data mart (17). Schaffner et al. (18)
In some developed countries, several hospitals proposed hub-and-spoke architecture and chose
have adopted a new type of logistics management centralized data warehouse as well as dependent
model, which is called SPD (supply, processing architecture of data mart. Wu et al. (19) suggested
and distribution) model (12). The SPD model the architecture of SOA-ITPA and adopted ser-
contains three links in the hospital logistics vice-oriented reusable component technology as
supply chain including supply, processing and well as extensive ETL (extract, transform and
distribution. In 1960s, Gordon A. Friesen, firstly load) services, in which the components commu-
put forward the idea of "hospital logistics man- nicate with each other via open standard message
agement and supply integration". The original protocol such as XML and SOAP. However,
purposes of the idea were to realize the integra- with the change of business environment, the
tion management of procurement, inventory, dis- existing BI systems can hardly meet the demands
tribution and consumption with the support of of efficient data analysis and collaborative busi-
informationization. With the recent development ness process optimization.
in this field, SPD model has been partly applied With the development of hospitals in terms of
to hospitals’ daily operation and management in construction scale, structure complication and
several countries including Japan and China (12- management informationization, the need for
15). In Japan, the SPD model has realized the scientific, standardized and lean management of
unified management of hospital logistics proce- hospitals is more obvious than ever. As one of
dures such as procurement, usage, recovery and the important aspects in hospital management,
distribution of medical products (e.g., drugs, logistics management is now facing the realistic
medical supplies and equipments) with the help problems of low efficiency and high cost.
of information systems. In China, several hospit- This study aimed to explore the key technologies
als in Shanghai, Nanjing, Tianjing and Lhasa, of hospital logistics collaborative BI system in-
have introduced the SPD model, in which the cluding architecture, data warehouse, data mining
procurement, inventory, package and distribution and collaboration, so as to improve the efficiency
of hospital supplies were entrusted to a third- of hospital SPD logistics management model.
party company, and hospitals only need to pay
for the supplies according to the actual consump- Materials and Methods
tion. Although there are many advantages of SPD
model in hospital logistics management, the intel- We performed literature review to understand
ligence and collaboration of SPD model are still SPD and BI related theories and recent research
relatively low, which hindered the improvement status. Firstly, we searched Engineering Village
of management efficiency to large extent.
database for English articles and China National based data preprocessing to construct the data
Knowledge Infrastructure (CNKI) for Chinese warehouse, we utilized improved data mining
articles published from 2011 to 2016. In addition, technologies including support vector machine
we searched Google for related books, confe- (SVM) supplier selection, supply chain collabora-
rence papers, Webpages, etc. The keywords for tion-based inventory optimization and swarm
literature review included "SPD", "supply, intelligence firefly algorithm-based multi-
processing and distribution", "BI", " business objective decision-making to solve key problems
intelligence", "hospital", "logistics", and "supply in hospital logistics collaborative BI system, we
chain". At last, we incorporated the ideas from used instant communication technology and SPD
the literature review into a practical study proto- workflow optimization method to realize the col-
col of the application of collaborative BI tech- laboration of BI system, and we selected Win-
nology in the hospital SPD logistics management dows Vista operating system as the development
model. environment of experimental platform. The
In this study, several technologies and methods overall research technology roadmap of study is
were employed. Specifically, we adopted the me- presented in Fig.1.
thods of rough set theory and firefly algorithm-
Fig. 1: The overall technology roadmap of the study (Note: BI=Business intelligence)
that the flow of information upstream was hete- the extra-preprocessed data for the use of the
rogeneous and the structure of raw data varied font-end. According to the relevant definition in
across different parts of supply chain, informa- the database, the data are named with clear busi-
tion systems, as well as modules, we took several ness meaning, i.e., integrate the tables, fields and
measures to define the metadata uniformly. At their complicated relationships into business
first, according to the Berlin core data set, we terms or indicator names that could directly be
provided metadata information for every member displayed in the font-end.
in the supply chain by reference to uniformed
metadata specifications that were defined by the Rough set theory and firefly algorithm-based
logistics alliance. Secondly, data exchange and data preprocessing
integration among various systems were achieved In order to improve the efficiency and accuracy
through metadata mapping. Finally, according to of swarm intelligence algorithm-based data min-
the idea of oriented to application integration ing, we used a rough set theory to perform data
(OAI), we designed a metadata framework and preprocessing of data warehouse. The data min-
information exchange model from the perspec- ing technology combined the rough set theory
tive of logistics information exchange to realize and firefly algorithm together, which not only
the data integration of different parts in the hos- took the quality of sub-data sets into considera-
pital logistics supply chain. tion, but also ensured the subtraction of raw data
Data storage refers to the storage mechanism of to the maximum extent. The objective function
shared data warehouse in hospital logistics colla- of the firefly algorithm is:
borative BI system. In order to solve the existing m X n R ( D)
Fitness ( X ) X ; m=|C|, refers to
contradiction between model stability and de- m m
mand variability, we stratified the data storage of the length of a firefly; (D) refers to the quality
X
shared data warehouse into three levels: (i) tem- of data sets; R is the subtraction of C (condition
porary buffer area, (ii) integrated data area and attribute). If the decision tables are continuous,
(iii) summary data area. The buffer function loads , or else .
1 2 1
multi-source system data in hospital logistics
supply chain into a temporary buffer area, the The basic steps are as follows: at first, we identi-
area works in the course of transform and load fied a proper method of subtraction (R), depend-
from data source system to integrated data area. ing on which firefly populations were randomly
Integrated data area, the core of the whole data generated by coding. Then each individual was
storage, evaluated by fitness function that has already
analyses every event from different dimensions been designed. At last, we figured out subsets
via constructing multi-dimensional model. The that met the function ( D) ( D) by iteration.
S c
integrated data area can be divided into three These steps helped us not only get the subsets
sub-areas: (i) operational data area, (ii) data analy- with the smallest data scale, but also keep the
sis area and (iii) data archiving area. The opera- original characters of data sets at the maximum
tional data area stores data that were classified extent.
according to the application themes and provides
uniformed hospital logistics data views based on Data mining technology research
real-time operation data. The data analysis area The data of collaborative BI system in the hos-
mainly stores historical data of hospital logistics pital SPD model is complex, multidimensional,
operation decision analysis. The data archiving and nonlinear. Using data mining, we were able
area is used to store the historical archived data to solve several problems including medical con-
derived from operational data area and data anal- sumables supplier selection, inventory optimiza-
ysis area. The summary data area is a virtual zone tion, as well as medical consumables distribution.
without actual stored data, it is arranged to store These were very important in the hospital logis-
tics management. In other words, by adopting tory optimization is considered a type of data-
improved data mining algorithm that met the driven analysis and design task, which requires
demand of collaborative BI system in the hospital analysis on large amounts of data followed by
SPD model, we achieved the following goals: (i) establishing a relational model between inventory
we realized more accurate and effective classifica- costs and service levels. According to the two-
tion and selection of medical consumables sup- stage inventory structure design, we divided the
pliers; (ii) we completed the procurement control, inventory optimization method into two steps: (i)
inventory optimization and inventory shortage we figured out the key parameters' approximate
warning in the first-level warehouse of the SPD fluctuation scopes of first-stage warehouse inven-
model; (iii) we solved the optimization problems tory in the SPD model using the time series me-
associated with medical consumables distribution thod; (ii) we designed discrete control parameters
in the secondary consumption points of the SPD and construct corresponding solution spaces, and
model, especially the problems associated with then identified the most feasible inventory opti-
the dynamic adjustment of distribution point, mization solutions by using swarm intelligence
period and amount. firefly algorithm.
that can be collaboratively supported in different efficiency, better control process, and effective
levels including information sharing support and management of logistics business processes.
task coordination support. Therefore, the busi- Therefore, we constructed the up and down
ness cooperation in the hospital SPD logistics streams and time constraint relations of workflow
model contains not only the information sharing by mining the time data from SPD workflow in
in traditional information system, but also the data warehouse, and then built the workflow
workflow optimization of information intelligent time constraint relations-based Activity On Edge
delivery and business-oriented collaboration. (AOE) Network. Based on which we figured out
the critical paths in SPD workflow net structure
Instant communication technology-based through the utilizing of swarm intelligence firefly
hospital logistics information collaboration algorithm and Critical Path Method (CPM).
Instant communication combines audio and vid-
eo communication and files transfer, and pro- Construction of experimental platform
vides necessary platform for information sharing This study intended to construct hospital logistics
and delivering of logistics collaborative BI sys- SPD model-oriented collaborative BI system, by
tem. Therefore, the full use of the instant com- which we realized the data and information shar-
munication technology requires necessary inte- ing between the hospitals, the suppliers and the
gration of heterogeneous information in the hos- manufacturers in the hospital logistics supply
pital logistics collaborative BI system. In this chain. By using SVM and swarm intelligence fire-
study, we provided uniformed structural descrip- fly algorithm methods, we analyzed the recent
tive data for describing information resources in years' basic information database of hospital in-
the hospital logistics collaborative BI system with formation system (HIS) and data provided by
the help of metadata. We accomplished this by companies in the logistics supply chain. We also
establishing metadata level of hospital logistics established a hospital logistics SPD model-
data warehouse using the Dublin core data sets, oriented collaborative BI system that realized
and built logistics information integration model. such roles as collaboration, data mining, data
We also attempted to build dynamic matching analysis, as well as application.
models of the supply chain and logistics related The timeliness and collaboration of hospital lo-
information (i.e., the information of demand, in- gistics collaborative BI system in SPD model
ventory, as well as distribution) in BI system via were improved by the following procedures: (i)
analyzing the attribute and operation information through constructing the module of sharing data
of the supply chain in the SPD model. warehouse, we realized the integration of data
marts in hospital logistics management, so as to
SPD workflow optimization-based hospital ensure the data safety and provide services such
logistics business collaboration as basic data process and analysis; (ii) for the ex-
Workflow is a part of Computer Supported Co- traction of inherent characteristics of data sets,
operative Work (CSCW), which plays an impor- we constructed the module of data preprocessing
tant role in researching how to make the groups to reduce the data dimension of collaborative BI
work cooperatively within the framework of in- system; (iii) by setting up supplier selection opti-
formationization. The SPD workflow is chiefly mization module and using the method of SVM,
used to describe the activities of tasks with fixed we improved the business processes to meet the
procedures in the hospital logistics businesses. demands of supplier evaluation, classification and
The core of the SPD workflow optimization is to selection in the hospital logistics SPD model-
figure out the "control points" that requires coo- oriented collaborative BI system; (iv) via estab-
perating by mining "up and down streams" and lishing inventory optimization module, we were
"time rules" in the original workflow. This is to able to solve the inventory optimization prob-
achieve several goals including improved work lems in hospital logistics warehouse; (v) by creat-
ing medical consumables constant distribution monitored by the logistics management depart-
module, we could deal with the decision-making ments, the inventory related variables can be
processes such as distribution amount, distribu- scientifically and coordinately arranged including
tion point, distribution lead time, as well as dy- the ordering point, safe inventory, maximum in-
namic adjustment of distribution amount; (vi) in ventory, and lead time of ordering. This way, we
order to realize the collaboration of logistics in- can significantly reduce the cost as well as the
formation and information sharing among every inventory risk. (iii) The distribution of consu-
parts of the supply chain, we built information mables is based on inventory optimization, by
collaborative service module; (vii) by building analyzing the historical consumption habit of
business collaborative service module, we pro- clinical departments, the distribution related va-
vided the decision-making basis for logistics riables including distribution amount, distribution
process optimization in the hospital SPD model. point, distribution lead time, as well as dynamic
In this study, we chose Windows Vista operating adjustment of distribution amount can be proper-
system as the development environment. Other ly assigned. By adopting this model, we may re-
specific requirements are as follows: software duce the waste of consumables and significantly
structure: J2EE; compile environment of Java reduce the workload associated with material
language: version JDK 1.5.07; server: open management and improve the efficiency of clini-
source server developed by Apache Group Jakar- cal activity.
ta group, version TOMCAT 5.0.28; JSP front- In China, the informationization of hospital
end development platform: Dreamweaver MX management mainly focuses on the information
2014; back-end development environment: Ec- support of the medical activity management, es-
lipse 5.1; database system: Microsoft SQL Server pecially the clinical treatment. The information
2008; ETL tools in BI system: PowerCenter; support for hospital logistics management is rela-
shared data warehouse: Microsoft SQL Server; tively weak, which makes it difficult to support
OLAP server: SAP Essbase; front display tools: the construction and implement of the SPD
Business Object. model. Specific problems were mainly manifested
in two aspects from the perspective of manage-
Discussion ment informationization. At first, the data utiliza-
tion rate was relatively low and the existing in-
With the rapid development of medicine and formation construction hardly met the data min-
health services and in the face of ever increasing ing demands in the complex logistics data. Addi-
demand, the use of medical consumables in hos- tionally, the collaboration between information
pitals is increasing. A hospital, usually, requires systems was quite lacking. Traditional construc-
multiple varieties of medical consumables with tion of informationization mainly concentrated
complicated specifications from several suppliers, on the information electronization in the hospit-
which results in considerable management diffi- al, rather than the outside links of the supply
culties, low management efficiency, as well as un- chain, which ignored the collaborative demands
safe medical activities. We advocate the use of among the hospitals, suppliers and manufactur-
the SPD logistics management model, which ers. With the development of recent decades, BI
have several advantages such as: (i) the procure- system has been widely applied in many fields
ment information is automatically generated by including manufacturing (20), bank (21), logistics
the system after analyzing the historical data on (22) and healthcare (23), etc. In the field of logis-
consumption, inventory level and response time tics, the application of BI system mainly focused
of suppliers. This may largely improve the stan- on the intelligent procurement management, in-
dardization and automation of procurement; (ii) telligent inventory management, as well as the
the inventory of every stage (e.g., the warehouse intelligent distribution schedule system (22, 24).
of center, departments and operating room) is While in the field of healthcare, BI system was
inventories. IIE Annual Conference. Proceedings. IEEE Workshop on Information and Software as
Institute of Industrial and Systems Engineers. Services (WISS), 52-9.
9. Costantino N, Dotoli M, Falagario M, Fanti MP, 19. Liya W, Barash G, Barash G, Bartolii C (2007). A
Mangini AM, Sciancalepore F, Ukovich W Service-oriented Architecture for Business In-
(2010). A model for the optimal design of the telligence. IEEE International Conference on Ser-
Hospital Drug Distribution Chain. Proceedings vice-Oriented Computing and Applications (SOCA
of the 2010 IEEE Workshop on Health Care '07), 279-85.
Management, 1-6. 20. Zhao XD, Zhang H, Lu JF (2010). Solution of
10. Dotoli M, Fanti MP (2008). A model to describe business intelligence application system
the hospital drug distribution system via first oriented to large iron & steel group. Computer
order hybrid Petri nets. 20th European Modeling Integrated Manufacturing Systems, 16(9):1999-
and Simulation Symposium, EMSS, 700-5. 2008.
11. Liu QX, Zhao Y (2012). Routing Optimization 21. Martin A., Manjula M. and Venkatesan P (2011).
for Hospital-internal Material Distribution. A Business Intelligence Model to Predict
Logistics Technol, 31(4):105-8. Bankruptcy using Financial Domain Ontolo-
12. Ito N, Chinzei M, Fujiwara H, Usui H, Hanaoka gy with Association Rule Mining Algo-
K, Saitoh E. (2006). Outline and effectiveness rithm[J]. Int J Compu Sci Issues, 8(3):1694-1703.
of support system in the surgical center by 22. Chen LY. Research on the Application of Busi-
supply, processing and distribution center ness Intelligence in Logistics Management
(SPD). Masui, 55 (4):478-85. and Decision-making Support [Master thesis].
13. Wei XY, Xiao MS, Wang H, Xiu Y (2015). College of Economics and Management,
Process Design and Application Practice of Nanjing University of Aeronautics and As-
the SPD System of Drugs. China Digital Medi- tronautics, China, 2006.
cine, 10(8):61-3. 23. Feng X, Yan SF (2011). Analysis on business in-
14. Ji M, Xi YM, Wu T (2013). Logistics manage- telligence application demand of hospital in
ment model of hospital drugs under the China. China Digital Medicine, 6(8):96-8.
background of new medical reform. Shanghai 24. Dai CL. The application of business intelligence
Medical & Pharmaceutical Journal, 34(19):25-7. in logistics industry [Master thesis]. College of
15. Dembińska-Cyran I (2005). Internal and external Software, East China Normal University,
supply chain of hospital. Log Forum,1(1):1-5. China, 2006.
16. Ariyachandra T, Watson HJ (2005). Key Factor 25. Liu B, Xia X, Chen YD, Zhu Y (2013). Research
in Selecting a Data Warehouse Architecture. of Hospital Decision Support System Based
Bus Intelligence J, 10(2):19-26. on Business Intelligence. China Medical Devices,
17. Ariyachandra T, Watson HJ (2006). Which Data 28(11):36-9.
Warehouse Architecture is Most Successful. 26. Zhang Z, Liu DY, Li J, Gao YF, Zhang Y, Chen
Bus Intelligence J, 11(1):4-6. ZQ (2011). Business intelligence-based hos-
18. Schaffner J, Jacobs D, Eckart B, Brunnert J, Zei- pital performance management system. Medi-
er A (2010). Towards Enterprise Software as cal Information, 24(2):32-3.
a Service in the Cloud. Proceedings of the 2nd