TY - GEN
T1 - An integrated data mining approach to real-time clinical monitoring and deterioration warning
AU - Mao, Yi
AU - Chen, Wenlin
AU - Chen, Yixin
AU - Lu, Chenyang
AU - Kollef, Marin
AU - Bailey, Thomas
PY - 2012
Y1 - 2012
N2 - Clinical study found that early detection and intervention are essential for preventing clinical deterioration in patients, for patients both in intensive care units (ICU) as well as in general wards but under real-time data sensing (RDS). In this paper, we develop an integrated data mining approach to give early deterioration warnings for patients under real-time monitoring in ICU and RDS. Existing work on mining real-time clinical data often focus on certain single vital sign and specific disease. In this paper, we consider an integrated data mining approach for general sudden deterioration warning. We synthesize a large feature set that includes first and second order time-series features, detrended fluctuation analysis (DFA), spectral analysis, approximative entropy, and cross-signal features. We then systematically apply and evaluate a series of established data mining methods, including forward feature selection, linear and nonlinear classification algorithms, and exploratory undersampling for class imbalance. An extensive empirical study is conducted on real patient data collected between 2001 and 2008 from a variety of ICUs. Results show the benefit of each of the proposed techniques, and the final integrated approach significantly improves the prediction quality. The proposed clinical warning system is currently under integration with the electronic medical record system at Barnes-Jewish Hospital in preparation for a clinical trial. This work represents a promising step toward general early clinical warning which has the potential to significantly improve the quality of patient care in hospitals.
AB - Clinical study found that early detection and intervention are essential for preventing clinical deterioration in patients, for patients both in intensive care units (ICU) as well as in general wards but under real-time data sensing (RDS). In this paper, we develop an integrated data mining approach to give early deterioration warnings for patients under real-time monitoring in ICU and RDS. Existing work on mining real-time clinical data often focus on certain single vital sign and specific disease. In this paper, we consider an integrated data mining approach for general sudden deterioration warning. We synthesize a large feature set that includes first and second order time-series features, detrended fluctuation analysis (DFA), spectral analysis, approximative entropy, and cross-signal features. We then systematically apply and evaluate a series of established data mining methods, including forward feature selection, linear and nonlinear classification algorithms, and exploratory undersampling for class imbalance. An extensive empirical study is conducted on real patient data collected between 2001 and 2008 from a variety of ICUs. Results show the benefit of each of the proposed techniques, and the final integrated approach significantly improves the prediction quality. The proposed clinical warning system is currently under integration with the electronic medical record system at Barnes-Jewish Hospital in preparation for a clinical trial. This work represents a promising step toward general early clinical warning which has the potential to significantly improve the quality of patient care in hospitals.
KW - deterioration warning
KW - feature selection
KW - real-time clinical monitoring
KW - time-series classification
UR - http://www.scopus.com/inward/record.url?scp=84866007817&partnerID=8YFLogxK
U2 - 10.1145/2339530.2339709
DO - 10.1145/2339530.2339709
M3 - Conference contribution
AN - SCOPUS:84866007817
SN - 9781450314626
T3 - Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining
SP - 1140
EP - 1148
BT - KDD'12 - 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining
T2 - 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD 2012
Y2 - 12 August 2012 through 16 August 2012
ER -