An Effective Information Support System in Medical Management: Indicator based Intelligent System Te-Lung Pan, Hsueh-Chi Shih

Abstract--Medical organization is a complex body composed of different units and process and this has unequivocally posed a dilemma in hospital management. In this study, we have summarized the requirements emerging from hospital’s indicators for medical management and proposed a new architecture that can effectively support it. The most important characteristics of this indicator base business intelligent system are enumerated as follows: 1) indicator data collection automation, 2) simplified BI’s KPI model by TQIP indicator project, 3) provides a convenient and effective graphical user interface and 4) supports outer feedback information rejoin into Data Warehouse for indicator comparison analysis. Index Terms-- Business Intelligent System, KPI, QIP

I. INTRODUCTION

M

edical organization is the combination of many different units and process and this has inevitably increased the difficulties of hospital management. Traditionally, hospital administrators or unit managers would use operation reports, for example: inpatients number, insurance amount or infected person as the essential data for their management decision. However, for the past recent years, pressures mainly attributed to the financial of Health Insurance system had brought about a huge management challenge to hospitals and therefore have to control cost more carefully but simultaneously need to maintain medical service quality. Competitions among hospitals make some hospitals survive and some not. The medical environment in Taiwan today is almost the same as in 70-80s’ America [7]. Most of hospitals adopt different information systems as an aid to strengthen hospital management and medical quality. And if we can collect data from different units in a hospital as well as inter-hospitals data; these gathered data by aggregation, will become a valuable management knowledge base for hospital decision maker [9]. However, there are two main difficulties that have to be tackled. First, how to construct a proper and complete decision

model or Key Performance Indicators (KPI) system. And secondly, how to implement an efficient framework that can collect data and share information between hospitals [2]. In this study we have presented an effective medical management Intelligent System that based on a QIP (Quality Indicator Project) model that can help hospital decision maker collect key indicators between hospitals and represent in a decision support way for them. II. INTELLIGENT SYSTEM AND DECISION MODEL Intelligent systems in general, are important components of an organization’s information systems [16] and senior managers believe intelligent systems can significantly contribute to organizational effectiveness and some organizations are strategically dependent on them [12]. There are two classes of intelligence tools which have been defined [5]. The first class is used to manipulate massive operational data and to extract essential business information from them like data mining. The second class is competitive intelligence aims at systematically collecting and analyzing information from the competitive environment to assist organization decision making [8] and is this study focused. Hospitals do not run in isolation. Therefore, running a hospital successfully does not just depend on an individual business basis, but rather how to run in comparison to others [15]. Fig. 1. shows the evolution that has taken place over the last few years. The initial step is to execute queries against operational data, resulting in reports or charts. The next logical step is to analyze the resulting data with traditional statistics or OLAP tools. We can also try to model the relationships in an organization (hospital) data to find out the behavior of organization under varying circumstances. In the last step, it is Business Intelligence (BI) in which knowledge about business is used to drive the decisions.

Te Lung Pan is a graduate student of information management department in National Yunlin Sciences and Technology University and chief of materials management department in St. Joseph’s Hospital at Taiwan. His research interests include the design and exploration of medical management support system ([email protected]). Hsueh-Chi Shih is a Professor of information management department in National Yunlin Sciences and Technology University. He earned his Ph. D. at University of Missouri-Rolla.

437

Medical Quality Indicators Project as Key Performance Indicators and will become the information definition or semantic views (Ontology) of this system. In this study we have adopted TQIP to be the KPI of medical intelligent system.

Fig. 1. Evolution from Queries to Business Intelligent

However, when Intelligent System or DSS (Decision Support System) have been developed some important problems are produced like: data definition, data transformation, and data redundancy and update problems. In order to avoid these problems nowadays, organizations need to have resolved the problem of sharing information between repositories and organization [2]. These problems may be partially eliminated by providing clear information definition or semantic views (Ontology) over different repositories and hide all the technical and organizational details associated with the data. Fig. 2. shows a general schema in which semantic views are used to integrate heterogeneous information content in the IS, DSS and other repositories.


A. Taiwan Quality Indicator Project There are a lot of medical quality indicator projects in the world and most of them have been developed in a systematic way recently. The most famous and longest-running of them is Quality Indicator Project (QIP) developed by Maryland Hospital Association (MHA) in 1985. In 1996 the Center for Performance Sciences (CPS) had independently launched and operated the International Quality Indicator Project (IQIP). QIP has demonstrated that performance measurements with indicators is not only possible but also bring significant results for the improvement of care [Kazandjian & Lied 1998]. In 1999, Taiwan Joint Commission on Hospital Accreditation (TJCHA) joined the IQIP as the local sponsor and added psychiatric long term care projects in 2001, establishing the current Taiwan Quality Indicator Project (TQIP). [6]. TJCHA’s utmost mission is hospital accreditation and healthcare quality promotion in Taiwan. That means there is strong link between accreditation and hospital’s medical quality. It thereby makes TQIP a must project for all hospitals in Taiwan. Table 1 shows the different medical indicator systems nowadays in the world. Indicato r system QIP

TQIP

HCFA

ACHS

Fig.2. A medical information definition example

III. KPI AND QIP IN MEDICAL ORGANIZATION MANAGEMENT To solve the obstacles and implement an efficient Business Intelligent System for decision support this study is proposing 438

Indicators characteristics 4 different pattern care settings: acute, psychiatric, long term and home. Classify by medical service pattern but only focus on medical service. Taiwan QIP, local sponsor of QIP. Same with QIP but home care is not included. Indicators for 5 conditions, acute W, pneumonia, CVA, a fib, CHF. Indicators were chosen based on efficacy evidence for the process of care. Australian Council on Healthcare Standards, 7 different sets of medical service indicators. The standard of clinical care standard developed by Australian Council on Healthcare Standards.

Main purpose Medical quality

Medical quality

Medical quality

Medical quality

JCAHO

THIS

Include clinical performance, health status, satisfaction perspective and administrative/financial functions. A unique grouping of performance measures to view a picture of the care provided in intensive care. Based on QIP and Donabedian evaluation model. An indigenous indicators developed by Taiwan College of Healthcare Executives.

collection is quite a heavy loading for execution and a suitable computer framework is necessary. Besides, after sending out one hospital’s own data to the organization of TQIP which will turn these data to IQIP (International QIP). The analyzed data which include other hospitals benchmark information all over the world who have joined this project will be sent back to the local hospital from IQIP. But how to apply these data on hospital management is another enormous and tedious task to accomplish. We can conclude here that the key problem for TQIP is the data sharing and application and the business intelligent information system can play an vital role and solve the problems that we have mentioned.

Medical quality and medical management performance

Medical quality and medical management performance

IV. PROTOTYPE ARCHITECTURE An architectural sketch for a complete medical indicator based BI system solution is proposed in Fig. 3.

Table. Ⅰ Medical indicator system in the world. [1, 23]

As of 2004, 74 hospitals were enrolled in the TQIP, including 17 medical centers (100%), 43 regional hospitals (62%), 6 district hospitals (14%) and 8 psychiatric hospitals (21%). It covers 65-70% of the national patient service volume. (Most district hospitals have joined the THIS program). B. Quality Indicator and its Difficult in Practice The practice of quality indicator project has a tremendous impact on accreditation and insurance payment asides from being a key tool for hospital administrator to control medical service quality. However, there are many difficulties have to be faced and most of these are computerization and the data collection. Table II are the top 5 obstacles during the practice of TQIP for hospitals. Obstructions Computerization Difficulty of data collection Difficulty of human resource insufficiency Indicators related units not fully supported Person unavailable for data collection Indicator definition unclear Others

Frequency Percentage 32 22 16

Percentage 72.7% 50.0% 36.4%

14

31.8% Fig. 3. A complete architecture for medical indicator based BI system

14

31.8%

8 3

18.2% 6.8%

Table. Ⅱ Data of top 5 obstacles to practice TQIP. [29]

Data collection of medical quality indicator can be classified into collection type and collection frequency. For example, employees and bed number of a hospital are stable number and the frequency for these indicators can be measured just annually; inpatient number special events will vary monthly or even daily. The frequency and the different type of data

The left side of the figure shows the data transformation architecture: Data Transformation Service extracts data from the operational data source and cleans/transforms/integrates data into Data Cube Transformation Service; data are then imported into a Data Warehouse by the formation of different data cube. Then the right side of the figure is the core component of this BI system – TQIP Based BIS which accessed data cube which comes from Data Warehouse and then provides this to the corresponding hospital manager. TQIP Based BIS also plays the role of transferring indicator data to QIP in XML form for further analysis. Data Cube Transformation Service can receive the analyzed information

439

coming from QIP and integrates it into hospital’s Data Warehouse. The main components in the medical indicator based BI are: z Data Transformation Service (DTS) integrates data from hospital operational databases; other data source and the XML file that feedbacks from QIP. The transformation rule and time period definition will be packed in COM objects. z Data Cube Transformation Service transforms integrate data into Data Warehouse in the form of data cube preparing for the usage of TQIP Based BIS. z Data Warehouse provides consistency, precision and long term storage for the usage TQIP Based BIS. z TQIP Based BIS is the core component for this Business Intelligent System which computes all the medical management indicators necessary at the different levels to provide graphical dashboard and grid reports. It also transforms part of indicators that need to be sent back to QIP in XML format. This medical indicators based BI architecture can analyzed TQIP indicators data through BI analyze rules to quantify the effectiveness of the strategy for decision maker. BIS can provide an alarm function allowing the unexpected events occurring at all levels and adopt the proper action. V. USER INTERFACE As sketched in Fig. 3., interactions with the user for BIS will be organized in different representation. Traditional report and OLAP will be presented and integrated with QIP indicators to give users a full picture of the trend of hospital in short and medium time trend (as figure a and figure b shows). Dashboard will also include QIP indicators but the information will be more real time in order to allow users to monitor the hospital real time information, for example the ICU bed occupation rate. And alarm will be quickly shown to enable users to react to the relevant events.

VI. SYSTEM DESIGN The prototype system uses the Indicator 4: Neonatal Mortality and all it’s sub indicators (4.1 – 4.8) for demonstration. Table 3 is the definition of TQIP indicator 4. Crate Business Data Model according to the definition of indicator and the Entities, Keys, Attributes and Relationships of source data will be the base of cube data. Create Dimensional Data Model for Indicator 4 this is the relation between dimensional data table and fact data table. Fig. 4. sketch the example of logical relation between data source, transformation and target data cube and Fig. 5. is the Data Transformation Service and it’s graphical design interface. By it’s Indicator 4: Neonatal Mortality 4.1 Neonatal Mortality for Direct Admissions, Birth Weight < 750g 4.2 Neonatal Mortality for Direct Admissions, Birth Weight 751 to 1,000g 4.3 Neonatal Mortality for Direct Admissions, Birth Weight 1,001 to 1,800g 4.4 Neonatal Mortality for Direct Admissions, Birth Weight > 1,801g 4.5 Neonatal Mortality for Transfers-in, Birth Weight < 750g 4.6 Neonatal Mortality for Transfers-in, Birth Weight 751 to 1,000g 4.7 Neonatal Mortality for Transfers-in, Birth Weight 1,001 to 1,800g 4.8 Neonatal Mortality for Transfers-in, Birth Weight > 1,801g Table. Ⅲ The Definition of TQIP Indicator 4.

Fig. a. Run chart

Fig. b. Grid report

Fig. 4. Logical relation between data source, transformation and target data cube.

440

[3]

[4]

[5] [6]

[7] [8]

[9]

Fig. 5. Data Transformation Service and it’s graphical design interface.

[10]

VII. EXPERIMENT AND EVALUATION We tested the prototype BI system with the methods below: System function test. We tested Data Transformation Service and Data Cube Transformation Service using the related data of Indicator 4. All the automation transformation service finished their job according to the predefined rule without any error. Integration test. By connecting every component of the prototype architecture to finish a Grid report and Dashboard presentation function. No obvious presentation delay was found. User experience test. This test was performed by multiple users to make sure the functions, user interface of this prototype system are satisfied with them. The result showed that the users were satisfied largely with the speed of information and graphical presentation and this system useful for provide a fast and effective method for medical indicators data collect and analysis and further application into medical decision support. However, this prototype system only presents indicator 4 and the users were doubtful with whether they can not continue using their familiar tools like Excel or some other statistical tools to analyze indicators data.

[11]

[12] [13]

[14] [15] [16] [17]

[18]

[19] [20] [21]

[22]

VIII. CONCLUSION

[23]

In this paper, we have summarized the requirements emerging from hospitals indicators for medical management and proposed a new architecture that can effectively support it. The most important characteristics of this indicator base business intelligent system are indicator data collection automation, simplified BI’s PKI model by TQIP indicator project, provide a convenient and effective graphical user interface and also supports outer feedback information rejoin into Data Warehouse for indicator comparison analysis.

[24] [25] [26]

[27]

[28] [29]

REFERENCES [1] [2]

ACHS, ACHS Clinical Indicator Summary Guide 2004, 2004. J. M. Blanco and A. Goni, “Building a Federated Database System: an Approach using a Knowledge Based System”, International Journal on Intelligent and Cooperative Information Systems, 3(4), 415-455, 1994.

441

R. Brachman, D. L. McGuinness, P. F. Patel-Schneider, L. A. Resnick and A. Borgida, “When and How to use a KL-ONE like Language, Principles of Semantic Networks”, pp. 401-456, 1990. R. Brachman and J. G. Schmolze, “An Overview of theKL-ONE Knowledge Representation System”, Cognitive Science, 9(2), pp171-216, 1985. Carvalho, R and Ferreira, M., “Using Information Technology to Support Knowledge Conversion Processes”, Information Research, vol. 7, 2001. Rong-Chi Chen, “Challenges Facing Taiwan in Assuring the Quality of Medical Care”, International Symposium: Toward on Equitable, Efficient, and High Quality National Health Insurance, Taipei, Taiwan, Mar. 18-19, 2005. Whe-Fang Chen, “The Analysis of Hospital Outsourcing in Taiwan Area”, Taiwan Public Health, Vol. 25, No. 4, pp. 233-244, 1999. W. Chung, H. Chen and Jr. J. F. Nunamaker, “Business Intelligence Explorer: A Knowledge Map Framework for Discovering Business Intelligence on the Web”, Proceedings of the 36th Hawaii International Conference on system Sciences. Victoria Hospital, Consistency of Triage in Victoria Emergency Departments, Education and Quality Report, July 2001. P. Devanbu and M. Jones, “The Use of Description logics in KBSE Systems”, Proceeding 17th International on Conference on Software Engineering, Sorrento, Italy, 1994. T. Gruber, “Toward Principles for the Design of Ontologies Used for Knowledge Sharing”, International Journal of Human and Computer Studies, 43(5/6), pp907-928, 1995. G. T., Gill, “Early Expert Systems: Where Are They Now”, MIS Quarterly (19:1), 1995, pp. 51-81. S. Gregor, “I., Explanations From Intelligent Systems: Theoretical Foundations and Implications for Practice”, MIS Quarterly (23:4), 1999, pp. 497-523. N. Guarino and P. Giaretta “Ontologies and Knowledge Bases: Toward a Terminological Clarification”, IOS, Press, Amsterdam, pp 25-32. J. Reinschmidt, H. Gottschalk, H. Kim and D. Zwietering, “Intelligent Miner for Data: Enhance Your Business Intelligence”, IBM, June 1999. F. Hayes-Roth, “Artificial Intelligence: What Works and What Doesn’t”, AI Magazine, Summer 1997, pp. 99-113. V. A. Kazandjian and T. R. Lied, “Caesarean section rates: effects of participation in a performance measurement project”, Joint Commission Journal of Quality Improvement 24, 187-196, 1998. V. A Kazandjian, N. Matthes and K. G. Wicker, “Are Performance Indicators Generic?” The International experience of the Quality Indicator Project, Journal of Evaluation in Clinical Practice, 9, 2, 256-272. G. Leroy and H. Chen, “Introduction to the special issue on decision support in medicine”, Decision Support systems, 2006. Maryland Hospital Association, “Quality Indicator Project Core Measures Component”, 2004. Miksch, S., Cheng, K., Hayes-Roth, B., An Intelligent Assistant for Patient Health Care, Autonomous Agents 97, ACM Press, Marina Del Ray, CA, 1997, pp. 458-465. R. H. Sprague and E. Carlson, Building Effective Decision Support Systems, Prentice-Hall, USA, 1982. S. Sprenger, Building a Framework for Comprehensive Indicator Measurement, ISQua, 2004. V. Dhar and R. Stein, Intelligent Decision Support Methods, The Science of Knowledge Work, PRENTICE HALL, 1997. E. Turban, Decision Support and Expert Systems, Management Support Systems, Prentice-Hall, 1995. E. Turban and D. Zhou, “A Group Decision Support Approach to Evaluating Journals”, Information and Management, Volume 42, Issue 1, 2004. L. A. Villamor and H. B. Calleja, “Quality Assessment research in the Management of Acute Myocardial Infarction in SLMC-Heart Institute: A Preliminary Report”, PJC Vol. 30 No. 2, April-June 2002. W. A., Woods and J. G. Schmolze, The KL-ONE Family. Semantic Networks in Artificial Intelligence, Pergamon Press, pp133-178, 1992. I. S. Xue, “The Anniversary Retrospection and Prospect of TQIP”, Conference of Taiwan Joint Commission on Hospital Accreditation, 2000.