Nonlinear Phenomena in Complex Systems, 2014, Vol.17, No.1, pp.50-56

NONLINEAR PHENOMENA IN COMPLEX SYSTEMS
An Interdisciplinary Journal

2014, Vol.17, No.1, pp.50-56

Analysis of 24-Hour Ambulatory Blood Pressure Monitoring Data Using Support Vector Machine
M. V. Voitikova, R.V. Khursa

This paper presents an effective hemodynamic classification algorithm for blood pressure (BP) monitoring data. The proposed approach takes into account two aspects of the hemodynamic states detection, namely the linear regression modeling of BP parameters and the classification block on the base of Data Mining algorithm called Support Vector Machine (SVM). At first, 4 features are extracted from the BP signals and then these features are reduced to only 2, finally, the SVM-classifier is used to classify the hemodynamic states. The proposed classification method is applied to clinical database. Thus 9 types of the hemodynamic states, including latent hypertension and high-risk hypertension, can be discriminated by SVM-classifier with the accuracy of 96%.

Key words: Data Mining, support vector machine, classification, regression, blood pressure, hemodynamics

Full text: Acrobat PDF (108KB)