Discriminant analysis of heart valve sounds during respiratory static volume conditions

Heart valve sounds (HVS), collected through a tracheal microphone, can provide a discriminant estimate of different respiratory static volume conditions. HVS data from 30 healthy male subjects were used to calculate descriptive statistical estimates of the HVS spectra, i.e. peak, bandwidth, and mean peak frequency. The HVS spectrum from DC to 50 Hz was divided into ten bands. The best six-variable model, for the classification of the respiratory static volume conditions, was obtained by an R/sup 2/ multivariable regression analysis. The overall classification accuracy of the discriminant program was 83%.<>