TY - GEN
T1 - Statistical Analysis of Inter-attribute Relationships in Unfractionated Heparin Injection Problems
AU - Wang, Haizhou
AU - Yang, Hui
N1 - Publisher Copyright:
© 2020 IEEE.
PY - 2020/7
Y1 - 2020/7
N2 - Unfractionated heparin (UFH) is commonly used in the intensive care unit (ICU) to prevent blood clotting. Recently, many researchers focus on the development of data- driven methods to solve UFH related problems, which usually involves time series analysis. The performance of data-driven methods depends on whether the inter-correlation of attributes (or variables) in the dataset is closely examined and addressed. This study performs attribute selection, optimal time delay and inter-attributes relations on ICU time series data, in order to provide insights of time series data for UFH related problems. Medical records of 3211 patients with 22 attributes extracted from MIMIC (Medical Information Mart for Intensive Care) III database are used for the experiment. Experimental result shows that some of commonly selected attributes in the literature are less sensitive to the variations of UFH injection. Furthermore, some attributes are inter-dependent, which can increase the complexity of data-driven models, implying that the number of attributes could be reduced. There are 9 attributes found highly related and fast responding in 22 commonly used attributes. This study shows strong potential to provide clinicians with information about sensitive attributes that can help determine the UFH injection policy in ICU.
AB - Unfractionated heparin (UFH) is commonly used in the intensive care unit (ICU) to prevent blood clotting. Recently, many researchers focus on the development of data- driven methods to solve UFH related problems, which usually involves time series analysis. The performance of data-driven methods depends on whether the inter-correlation of attributes (or variables) in the dataset is closely examined and addressed. This study performs attribute selection, optimal time delay and inter-attributes relations on ICU time series data, in order to provide insights of time series data for UFH related problems. Medical records of 3211 patients with 22 attributes extracted from MIMIC (Medical Information Mart for Intensive Care) III database are used for the experiment. Experimental result shows that some of commonly selected attributes in the literature are less sensitive to the variations of UFH injection. Furthermore, some attributes are inter-dependent, which can increase the complexity of data-driven models, implying that the number of attributes could be reduced. There are 9 attributes found highly related and fast responding in 22 commonly used attributes. This study shows strong potential to provide clinicians with information about sensitive attributes that can help determine the UFH injection policy in ICU.
UR - http://www.scopus.com/inward/record.url?scp=85091045596&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85091045596&partnerID=8YFLogxK
U2 - 10.1109/EMBC44109.2020.9176645
DO - 10.1109/EMBC44109.2020.9176645
M3 - Conference contribution
C2 - 33019196
AN - SCOPUS:85091045596
T3 - Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS
SP - 5374
EP - 5377
BT - 42nd Annual International Conferences of the IEEE Engineering in Medicine and Biology Society
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 42nd Annual International Conferences of the IEEE Engineering in Medicine and Biology Society, EMBC 2020
Y2 - 20 July 2020 through 24 July 2020
ER -