Abstract:
The problem of constructing a compact mathematical model representing various fragments of electroencephalograms (EEGs) during epileptic absence seizures is considered. Such a model would be useful for solving a number of practical tasks related to the clustering of time series, seeking for relationships between various leads, separating norm and pathology, etc. It is shown that an adequate model that takes into account the EEG structure can be constructed using nonuniform embedding. The dimensionality, degree of nonlinearity, and lags are selected based on an objective numerical criterion.
Fulltext:
PDF file (1131 kB)
