Alarm Fatigue Analysis
Extracting Patterns in ICU alarm logs at UIUC
July - October 2014
I worked on this project as a student researcher for the UIUC DEPEND group during the Summer and Fall of 2014. Our group collaborated with Rush University Medical Center in Chicago’s west loop to obtain alarm log data from ICU patient monitors for further analysis.
Problem
The overabundance of false alarms from ICU Patient Monitoring Systems impedes proper care-taking in hospitals, and actually increases patient risk. Doctors, nurses, and other hospital workers become desensitized to constant alarms and respond slower, if at all, to these warnings. This can result in further patient illness or death when significant physiological events occur, but are not monitored.
Alarm Hazards rank first on the Emergency Care Research Institute's list of Top Ten Health Technology Hazards for 2012
Goal
Through alarm analysis:
Project
 Coding in RI was responsible for developing R scripts that would extract data such as alarm type, frequency, duration and correlation to physiological events from the alarm output logs from the Rush ICU |  IRB ProposalI wrote and submitted a successful IRB proposal at Rush to grant the team access to patient monitor logs and physiological waveform data to allow for the study to continue beyond my time there. | Meeting with Medical ProfessionalsI presented preliminary research to the CTO of Rush to secure quicker and easier access to the data the team needed. |
|---|
Create better algorithms to help mitigate the incidence of false alarms in monitoring systems
Correlate this data to physiological events
Glean insights into which alarms are ignored
Discover how long each alarm is left on before silencing
Discover how many alarms go off, and what type