Byungchae Lee

Using Neural Network to Recognize Human Emotions from Heart Rate Variability and Skin Resistance

The purpose of this paper is to estimate emotions using neural network and the changes in activities of autonomic nervous system (ANS). Since ANS cannot be controlled artificially, we presumed that the changes in emotions would be reflected to the changes in ANS. In order to observe those changes, we provided the subjects with some video clips which can induce a variety of emotions and measured the changes in ANS, especially in heart rate variability (HRV) and in galvanic skin response (GSR). With those analyzed results from the experiments, we established an algorithm based on neural network, finally we could reach the estimating rate of 80.2%

An Algorithyn For Pattern Recognition Of Multichannel ECG Signals

This paper describes an algorithm that can efficiently analyze the multichannel ECG signal using the frame. The input is a set of significant f eatures( points) which have been extracted from an original sampled signal by using the split-and-merge algorithm. A signal from each channel c an be hierarchically decomposed to produce a hierarchical AND/OR graph on the b asis of the priori knowledge for ECG signal. The search mechanism with some heuristics and the mixed paradigms of data-driven and model-driven hypothesis f ormation are used as the major control mechanisms. The mutual relations among features are a lso considered by evaluating a score based on the r elational spectrum. For recognition of morphologies c orresponding to OR nodes, an hypothesis m odification strategy is used. Other techniques such as instance inheritance, priority update of prototypes, and template m atching facility are also used. exactly recognized the primary points and supporting points from the multichannel ECG signa 1 s. This a lgorithm

Design of portalble 12 leads automated ECG analyzer using 68000 MPU

This paper describes the hardware/software design of an automated ECG analyzer in order to precisely measure and analyze waveforms of standard 12 leads for classification of disease. The analysis results and ECG waveforms are printed on the thermal printer. Under this hardware framework the peak detection and diagnosis programs are implemented using pattern recognition and Minnesota diagnosis criteria. Especially this paper focuses on the prototype system hardware which can diagnose the heart desease without medical experts helping and can involve various diagnosis algorithms.

A complexity measure of heart rate variability in time domain

Describes a complexity measure algorithm based on nonlinear dynamics (chaos theory). We have proposed a fractal dimension-1 and fractal dimension-2 algorithm with a digital filter for quantifying complexity or regularity of the biomedical signal. The approximate entropy algorithm, which measures the system regularity, is also compared. These quantified complexity measures may provide useful information about human physiology.