Detecting Anomalies in Alert Firing within Clinical Decision Support Systems using Anomaly/Outlier Detection Techniques
Soumi Ray, Adam T. Wright · 2016
Clinical Decision Support (CDS) systems play an integral role in the improvement of health care quality and safety. Alert malfunctions within CDS are a common problem and these greatly limit its usability. Anomaly detection is a novel approach to identify malfunctioning within CDS systems. Once an anomaly in alert firing is detected, it can be used to rectify and potentially fix the CDS system to make it robust and reliable. We introduce and apply four anomaly detection algorithms to estimate the dates when malfunctions and/or changes in alert firing occur. Preliminary results demonstrate successful detection of anomaly occurrences within the alert data; this carries the potential to be used for root cause analysis of such malfunctions.