Machine Learning
LIBSVM: A library for support vector machines
ACM Transactions on Intelligent Systems and Technology (TIST)
Hi-index | 0.00 |
In this study, we investigate the feasibility of using the linear cross-correlation function (CCF) and nonlinear correlation dimension (CD) features of mean arterial blood pressure (MABP) and mean cerebral blood flow velocity (MCBFV) as a signature to classify diabetics with various degrees of autonomic neuropathy. 72 subjects were recruited. For each subject, continuous CBFV was measured using a Transcranial Doppler ultrasound, and continuous ABP recorded using a Finapres device in supine position. The CCFs and CDs of pre-filtered spontaneous MABP and MCBFV were computed. Twelve CCF features and one CD feature were extracted from three frequency ranges: very low frequency (VLF, 0.015-0.07Hz), low frequency (LF, 0.07-0.15 Hz), and high frequency (HF, 0.15-0.40Hz). The feature vectors are classified using a support vector machine (SVM) classifier; and a classification rate of 91.67% is obtained under the leave-one-out cross validation evaluation scheme. This very encouraging result indicates that the proposed linear and nonlinear features between MABP and MCBFV can be effective features to discriminate diabetic patients with autonomic neuropathy.