The method evaluation for preictal prediction of epilepsy with strong-noise EEG and simulation of automatic drug release system

摘要

Epilepsy is a disease with chronic disorder of nervous system, which owns the characteristic of repeat attacking and difficult to cure. So, it's significant to study the forecasting method and give preventive treatment. In this paper, the approximate entropy and the maximum Lyapunov exponent based on the improved Wolf algorithm were used to extract and analyze the characteristic of nonlinear dynamics of the human epileptic electroencephalograph under different states, the results indicated that two methods can effectively predict preictal state and approximate entropy show better performance. Results calculated in the wavelet domain were compared between the two methods and better effect can be obtained under first order discrete wavelet transform. The early waning and automatic drug release system was simulated by LabVIEW Virtual Instrument, with 80% sensitivity and 90.9% specificity. The automatic process and drug-release rate controlling by characteristic value is simulated, which show important application value.