• DocumentCode
    1369507
  • Title

    Building the IMPROVE data library

  • Author

    Korhonen, Ilkka ; Ojaniemi, Jyrki ; Niieminen, K. ; Van Gils, Mark ; Heikelä, Arno ; Kari, Aarno

  • Author_Institution
    VTT Inf. Technol., Tampere, Finland
  • Volume
    16
  • Issue
    6
  • fYear
    1997
  • Firstpage
    25
  • Lastpage
    32
  • Abstract
    Well-characterized and comprehensive test data sets are essential in the development and evaluation of biosignal interpretation methods for intensive-care patient monitoring. The IMPROVE (IMPROVing control of patient status in critical carE) data library is an annotated data library that contains practically all the monitored and other clinical data from critically ill patients at high risk of oxygen-transport-related problems. The data, collected at the Intensive Care Unit of the Kuopio University Hospital, is supported by continuous patient-state assessments carried out by a bedside physician and includes 59 patient records, each having a typical duration of 24 hours. In this article, we describe the technical setup of the system, including what signals and parameters were collected and how the annotations were done. We also discuss the lessons learned from the data collection process
  • Keywords
    data acquisition; medical information systems; medical signal processing; patient care; patient monitoring; IMPROVE data library; Kuopio University Hospital; annotated data library; annotation; biosignal interpretation methods; clinical data; continuous patient-state assessments; critical care; critically ill patients; data collection process; intensive-care patient monitoring; oxygen-transport-related problems; patient records; patient status control; technical setup; test data sets; Biomedical engineering; Biomedical monitoring; Costs; Databases; Hospitals; Laboratories; Libraries; Medical tests; Patient monitoring; System testing;
  • fLanguage
    English
  • Journal_Title
    Engineering in Medicine and Biology Magazine, IEEE
  • Publisher
    ieee
  • ISSN
    0739-5175
  • Type

    jour

  • DOI
    10.1109/51.637114
  • Filename
    637114