[Improving data warehouse environments for efficient analysis of long time-series data].

Medical records contain enormous amounts of data. It is important to extract useful evidence from such data and feedback to clinical medicine. Evidence-based medicine (EBM) was introduced in the 1990s and has been widely used for more than 20 years, however, hospital information system environments that take advantage of the ideas of EBM have not yet been established. Recently, the numbers of medical institutions with multilateral search systems for the medical records stored in data warehouses (DWHs) have been increasing, but these institutions' systems cannot deal fully with issues such as data reliability and high-dimensional, high-speed searches. DWHs can control long time-series data. Although, the measurement methods and analytical equipment used have been modified and improved with advances in testing techniques, this may have induced shifting and/or fragmentation of these types of data. Furthermore, database design has to be flexible to satisfy the various demands of information retrieval; systems must therefore have the structures to deal with such demands. We report here our new system infrastructure, which exchanges data in order to absorb the data shifting associated with changes in the testing methods. The system enables the preparation of DWH environments that can be used to seamlessly analyze long time-series data, record in knowledge databases the results of comprehensive analyses of institutions' characteristics of laboratory diagnoses, and use the data in education, research and clinical practice.