Premium
A Data‐Based Method for Bivariate Outlier Detection: Application to Automatic Blood Pressure Recording Devices
Author(s) -
Clark L.A.,
Denby L.,
Pregibon D.,
Harshfield G.A.,
Pickering T.G.,
Blank S.,
Laragh J.H.
Publication year - 1987
Publication title -
psychophysiology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.661
H-Index - 156
eISSN - 1469-8986
pISSN - 0048-5772
DOI - 10.1111/j.1469-8986.1987.tb01872.x
Subject(s) - bivariate analysis , outlier , covariate , homogeneous , psychology , blood pressure , anomaly detection , variety (cybernetics) , statistics , data mining , computer science , artificial intelligence , medicine , mathematics , combinatorics
The rapidly increasing use of automatic devices for the measurement of blood pressure has made it increasingly important to identify artifactual readings. This paper describes an objective technique for isolating a small but adjustable percentage of the readings that are likely artifacts. The validity of these readings requires the subjective judgment of the clinician. The proposed methodology substantially reduces the number of readings to be manually examined by not only taking into account the level of systolic and diastolic blood pressure per se, but also the relationship between the two and how they vary according to differential covariate information. The method can be applied to homogeneous populations, or as illustrated in this paper, to heterogeneous populations after adjustment for known or suspected sources of variability. Thus the technique is applicable to protocols which examine blood pressure, or for that matter any two (or more) related variables, during a variety of experimental procedures including psychophysiological reactivity tasks.