A wavelet multiresolution and neural network system for BCG signal analysis

Ballistocardiography (BCG) has an interesting feature that no electrodes are needed to be attached to the body during measurements. This provides a potential application to assess the patients heart condition in the home. Artificial neural networks (ANNs) have several properties that make them promising for the automatic signal classification problems. In the time domain of the BCG classification, the whole cardiac cycle of BCG waveform needs a large size neural network and a large training sample which make the classification a computationally intensive task. By classifying the data in a compressed format, savings in computer time may be realised. In this paper, we used wavelet multiresolution analysis that allows significant information content of the BCG signal to be obtained. Small subsets of the wavelet coefficients were used to classify the normal hypertension and heart attack risk subjects by a single hidden layer neural network. It is shown that the proposed system achieved overall 94.66% correct classification rate for testing the data set. The advantage of the proposed classification system is to reduce the computation complexity and to be easily implemented into a standalone device for real time application