A seizure prediction algorithm is proposed that combines novel multivariate EEG features with patient-specific machine learning. The algorithm computes the eigenspectra of space-delay correlation and covariance matrices from 15-s blocks of EEG data at multiple delay scales. The principal components of these features are used to classify the patient's preictal or interictal state. This is done using a support vector machine (SVM), whose outputs are averaged using a running 15-minute window to obtain a final prediction score. The algorithm was tested on 19 of 21 patients in the Freiburg EEG data set who had three or more seizures, predicting 71 of 83 seizures, with 15 false predictions and 13.8 h in seizure warning during 448.3 h of interictal data. The proposed algorithm scales with the number of available EEG signals by discovering the variations in correlation structure among any given set of signals that correlate with seizure risk.

The problem addressed in this study is how to design and test compact antenna arrays for portable Mulitple-Input Multiple-Output (MIMO) transceivers. Mutual coupling in an antenna array affects signal correlation and array radiation efficiency -both of which have dramatic consequences for MIMO channel capacity.Mutual coupling becomes more pronounced as array aperture shrinks and is therefore a critical issue in compact array design. Two novel compact arrays are designed and fabricated for use in MIMO enabled mobile devices.These arrays are extremely compact yet demonstrate acceptable mutual coupling and radiation efficiency because of the MIMO-specific criteria used during their design. An experimental methodology is presented for fair and meaningful characterization of MIMO arrays by field trial. This methodology addresses the issue of capacity normalization and quantifies how well an antenna array's radiation pattern interfaces with multipath propagation.Results are presented from an extensive measurement campaign in which a true MIMO transceiver testbed is outfitted with compact arrays and dipole arrays of various sizes. A comprehensive and fair comparison is made between the compact arrays and dipole arrays in a variety of indoor propagation scenarios. Design recommendations for compact MIMO arrays are given.

A patient-specific seizure prediction algorithm is proposed that extracts novel multivariate signal coherence features from ECoG recordings and classifies a patient's pre-seizure state. The algorithm uses space-delay correlation and covariance matrices at several delay scales to extract the spatiotemporal correlation structure from multichannel ECoG signals. Eigenspectra and amplitude features are extracted from the correlation and covariance matrices, followed by dimensionality reduction using principal components analysis, classification using a support vector machine, and temporal integration to produce a seizure prediction score. Evaluation on the Freiburg EEG database produced a sensitivity of 90.8% and false positive rate of .094.

Apnea of prematurity is a common developmental disorder in preterm infants that is implicated in a number of acute and long-term complications. Therapeutic stochastic resonance (TSR) is a noninvasive preventative intervention for stabilizing breathing patterns and reducing the incidence of apnea and hypoxia. Because the stabilizing effect of TSR lags its initiation, it can be used most effectively if it is linked to a system for apnea prediction. We present a real-time algorithm for generating apnea predictions based on cardio-respiratory and movement features extracted from multiple physiological sensors. The features are used to create patient-specific statistical models of apnea precursors. The state parameters generated by these models are evaluated over time to form apnea predictions. The algorithms predictions are evaluated using a short, 5.5 minute prediction horizon. The algorithm obtains highly accurate predictions, with statistical significance obtained on five out of the six patients that it is evaluated on.