TY - GEN
T1 - Automating the identification of patient safety incident reports using multi-label classification
AU - Wang, Ying
AU - Coiera, Enrico
AU - Runciman, William
AU - Magrabi, Farah
N1 - Copyright the International Medical Informatics Association (IMIA) and IOS Press 2017. Version archived for private and non-commercial use with the permission of the author/s and according to publisher conditions. For further rights please contact the publisher.
PY - 2017
Y1 - 2017
N2 - Automated identification provides an efficient way to categorize patient safety incidents. Previous studies have focused on identifying single incident types relating to a specific patient safety problem, e.g., clinical handover. In reality, there are multiple types of incidents reflecting the breadth of patient safety problems and a single report may describe multiple problems, i.e., it can be assigned multiple type labels. This study evaluated the abilty of multi-label classification methods to identify multiple incident types in single reports. Three multi-label methods were evaluated: binary relevance, classifier chains and ensemble of classifier chains. We found that an ensemble of classifier chains was the most effective method using binary Support Vector Machines with radial basis function kernel and bag-of-words feature extraction, performing equally well on balanced and stratified datasets, (F-score: 73.7% vs. 74.7%). Classifiers were able to identify six common incident types: falls, medications, pressure injury, aggression, documentation problems and others.
AB - Automated identification provides an efficient way to categorize patient safety incidents. Previous studies have focused on identifying single incident types relating to a specific patient safety problem, e.g., clinical handover. In reality, there are multiple types of incidents reflecting the breadth of patient safety problems and a single report may describe multiple problems, i.e., it can be assigned multiple type labels. This study evaluated the abilty of multi-label classification methods to identify multiple incident types in single reports. Three multi-label methods were evaluated: binary relevance, classifier chains and ensemble of classifier chains. We found that an ensemble of classifier chains was the most effective method using binary Support Vector Machines with radial basis function kernel and bag-of-words feature extraction, performing equally well on balanced and stratified datasets, (F-score: 73.7% vs. 74.7%). Classifiers were able to identify six common incident types: falls, medications, pressure injury, aggression, documentation problems and others.
KW - Machine learning
KW - Patient safety
KW - Risk management
UR - http://www.scopus.com/inward/record.url?scp=85040512077&partnerID=8YFLogxK
U2 - 10.3233/978-1-61499-830-3-609
DO - 10.3233/978-1-61499-830-3-609
M3 - Conference proceeding contribution
AN - SCOPUS:85040512077
SN - 9781614998297
VL - 245
T3 - Studies in Health Technology and Informatics
SP - 609
EP - 613
BT - MEDINFO 2017
A2 - Gundlapalli, Adi V.
A2 - Jaulent, Marie-Christine
A2 - Zhao, Dongsheng
PB - IOS Press
CY - Amsterdam
T2 - 16th World Congress of Medical and Health Informatics: Precision Healthcare through Informatics, MedInfo 2017
Y2 - 21 August 2017 through 25 August 2017
ER -