A TIME-SERIES PRE-PROCESSING METHODOLOGY FOR BIOSIGNAL CLASSIFICATION USING STATISTICAL FEATURE EXTRACTION

摘要: Biosignal classification is an important diagnosis tool in biomedical application that helps medical experts to automatically classify whether a sample of biosignal under test/monitor belongs the normal type or otherwise. Most biosignals are stochastic and nonstationary nature, means their values timedependent statistics vary over different points time. However, most algorithms data mining designed work with possess multiple attributes order capture non-linear relationships between predicted target class. Therefore it has been crucial research topic for transforming univariate time-series multivariate dataset fit into algorithms. For this, we propose pre-processing methodology, called Statistical Feature Extraction (SFX). Using SFX can faithfully remodel statistical characteristics via sequence piecewise transform functions. The new methodology tested through simulation experiments three representative types biosignals, namely EEG, ECG EMG. yield encouraging results supporting fact indeed produces better performance than traditional analyses techniques like Wavelets LPC-CC.