Design an Intelligent Ballistocardiographic Chair using Novel QuickLearn and SF-ART Algorithms and Biorthogonal Wavelets

Abstract

To design a heart diseases diagnosing system, we applied compactly supported biorthogonal wavelet transform to extract essential features of the ballistocardiogram (BCG) signal and to classify them using two novel supervised learning algorithms called SF-ART and quicklearn. Initial tests with BCG from six subjects (both healthy and unhealthy people) indicate that both SF-ART and quicklearn algorithms can classify the subjects into three classes with high accuracies, high learning speeds, and very low computational loads compared to the well-known neural networks such as multilayer perceptrons. The proposed heart diseases diagnosing systems are almost insensitive to latency and nonlinear disturbance. Moreover, the wavelet transform requires no prior knowledge of the statistical distribution of data samples and the computational complexity and training time are reduced.