Competence Center for Medical Data Warehousing and Analysis

The Competence Center for Medical Data Warehousing and Analysis (MEDAN) will be established as a four-year competence center in the Database and Information Systems (DIS) group in the Faculty of Computer Science at the Free University of Bozen-Bolzano (FUB). The creation of this competence center has been triggered by a strategic project at Hospital Meran (HM) with the aim of realizing a data warehouse that forms the backbone of a data analysis platform that supports the development of value added services for the health district South Tyrol. The center’s mission is to conduct basic and applied research in the area of data warehousing and intelligent data analysis for health care and medical applications. The competence center is closely collaborating with international research institutions to ensure high quality research. It works with real-world data and application problems from HM to foster a successful know-how and technology transfer to health care institutions in South Tyrol and to provide targeted scientific support for the data warehouse project at HM. 1. BACKGROUND AND MOTIVATION Data Warehousing and Data Mining Data warehouses emerged in the late 1980s and early 1990s and were designed to meet the increasing demand to facilitate the extraction and reporting of information from ever growing amounts of data [17, 18]. Traditional operational systems were unable to meet this need for a number of reasons: the processing load of reporting reduced the response time of the operational systems; the database designs of operational systems were not optimized for information analysis and reporting; the presence of different operational systems significantly complicates the company-wide reporting; and the development of reports in operational systems often required writing specific programs, which is slow and expensive. Data warehouses were specifically designed to support the management of data coming from different data sources, and to analyze and report this data. The construction of a data warehouse in combination with the use of advanced data analysis tools to gain accurate, consistent and fast access to important information that supports strategic decisions is a key to gain competitive advantages. This also holds for the health care sector, which is characterized by huge amounts of complex and heterogeneous data, e.g., semi-structured data, medical images, numerical and symbolic data, temporal data, data expressed at different scales, etc. Extracting information from such data puts forward a number of challenges and calls for advanced data warehouse and data mining technologies. The ability to effectively use state-of-the-art data warehouse and data mining technologies is essential to ensure quality, success and competitiveness of health care organizations. Thus, it is of strategic regional importance to establish and maintain a solid and up-to-date competence in this area and to build a sustainable collaboration between the university and the health care institutions in the province. A Data Warehouse at the Hospital Meran Administrative Data Warehouse The ICT department at HM experiences an increased need to build a data warehouse and analysis system to satisfy the demands of various user groups, including decision makers, physicians and administrators. All user groups need quick and easy access to reliable information that is of strategic importance. More specifically, the controlling department has the following duties: • Systematic provision of basic data and information to decision makers, including the responsible for the budget. Such data have to be extracted from the Hospital Information System (HIS) and from other distributed sources (e.g., from the finance and personnel departments), including external institutions (e.g., Martinsbrunn and St. Anna). • Ad hoc provision of data and information in response to specific requests from the provincial government. Typically such requests require that the data are extracted and collected in an ad hoc manner. • Elaboration of the internal budget and business plans, considering different heterogeneous budget centers, and the elaboration of business plans. These tasks require as input detailed information that can be obtained by an analysis of historical data. The extraction, analysis, and preparation of this data is a time consuming and error-prone process that is mainly based on spreadsheet tools. The situation is further complicated because some of the data are duplicated and maintained centrally in the health care department of the provincial government. This situation not only leads to redundant data storage and additional maintenance costs, but the data easily get inconsistent and yield erroneous analysis results. To cope with these problems, HM started a project with the aim to develop a data warehouse and data mining system. The relevant data shall be imported from the various operational systems and stored in a central data warehouse (see Figure 1). The main characteristics of the data warehouse approach are: the data storage is not application-oriented as in the operational systems; data from different applications will be merged; historical health care information is systematically handled; and the redundant storage in the data warehouse reduces the workload of operational systems. Figure 1: Data Sources for the Data Warehouse. For the development and maintenance of the data warehouse the following aspects are crucial: the consistency of the imported data; the automatic import of the data in order to minimize the maintenance overhead for the data warehouse; the scalability of the data warehouse system such that new requirements can be implemented without huge efforts; a high level of detail in the data to allow flexible drill-down analysis; the relevance of the collected data for the improvement of strategic processes of the hospital and the support of critical decision making activities; and well defined and simple-to-use interfaces that can be accessed by data mining tools without additional programming efforts. The data warehouse provides a central repository of consistent and reliable data that can be accessed by various analysis and mining tools to support decision makers in performing detailed and complex business analysis in an easy and flexible way as illustrated in Figure 2. Figure 2: Data Warehouse and Analysis Solution. Medical Data Warehouse While the first part concerns mainly the collection and analysis of administrative and financial data, a second step aims to extend the system with a medical data warehouse that collects clinical data sources. The promise of this integration is a better understanding of the financial and organizational implications of clinical decisions such as tests or treatment options in the care of patients. For example, specific clinical practices and procedures might turn out to be less expensive but produce the same (or even better) clinical outcome. A medical data warehouse in combination with data mining technologies provides also a useful resource to support evidence based medicine, i.e., to make clinical decisions about the care of individual patients using the best available evidence. Methods in association rule mining can be applied, e.g., to discover rare patterns, thus generating clinical evidence to support clinicians to take more informed decisions. While administrative and financial data are characterized by a regular structure, capturing clinical data turns out to be more complicated, since medical data are much more complex and heterogeneous: semi-structured and diversely structured data, medical images, numerical and symbolic data, temporal data, data expressed in different scales and granularities, etc. Similar as for the modeling and the storage of the data, the analysis of clinical data possibly in combination with administrative data poses new challenges due to the increased amount, dimensionality, and diversity of the data. 2. MISSION AND OBJECTIVES The competence center has the mission to create competences in data warehouse and data analysis technologies and to contribute towards increasing the critical awareness in these areas in the health care sector in South Tyrol in the near and longer terms. This shall be achieved by conducting high-quality research at an international level and by investigating problems that arise from applications in local health care institutions. The competence center will progress current data warehouse and analysis technologies in health care applications. Substantial resources will be invested into sustainable basic research activities in collaboration with leading international researchers. The center’s mission includes also the transfer of know-how and technology from research to practice, which includes the implementation and evaluation of prototype systems in real-world settings given by HM. The research activities of the competence center are embedded into a strategic project at HM, which foresees the development of a data warehouse and analysis system that later shall be extended to the entire health care district South Tyrol. The project at HM not only puts forward real-world user requirements and problems that require further investigations, but also provides a testbed to evaluate the developed technologies with large volumes of real-world data.