Published February 24, 2021 | Version v1.1.1
Journal article Open

Resources for a "Do-it-Yourself Analysis" of the Patient Monitoring Alarm Data from Intensive Care Units

  • 1. Department of Anesthesiology and Intensive Care Medicine, Charité – Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institute of Health, Berlin, Germany; Einstein Center Digital Future, Berlin, Germany
  • 2. Department of Anesthesiology and Intensive Care Medicine, Charité – Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institute of Health, Berlin, Germany
  • 3. FG Arbeitswissenschaft, Technische Universität Berlin, Berlin Germany

Description

The clinical audit logs were manually collected from the patient monitoring system of a 21-bed intensive care unit (ICU) from a large German hospital via USB stick from the central patient monitoring device three times during 2019 (in winter, summer and autumn). The data consists of the time, bed number, alarm type (i.e., parameter, device, alarm criticality) and alarm handling (e.g., threshold adjustments, use of the pause function). In total, data span 93 calendar days. No actual patient identifying data elements were collected. For further deidentification, dates were shifted into the future by a pseudo-random offset for all patients; the bed number was replaced by a pseudonym. Day and night rhythm, weekends, the season and the bed characteristic (double room, single room) were not affected by this process.

We provide the fully annotated R scripts that we used to conduct the alarm data analysis to enable even beginners in R to do likewise. Further explanations can be found in the results section of the accompanying publication and in the scripts. The R-Markdown file can be used to create comprehensive alarm reports.

Files

data-science-charite/ICU_DIY_AlarmAnalysis-v1.1.1.zip

Files (26.4 MB)

Additional details