Аннотация
An optimal algorithm for classification of respiratory sound on the basis of spectral-correlation criteria is analyzed and parameters of its effectiveness are determined. These parameters are analyzed numerically of for an idealized model of the human respiratory tract. The comparative diagnostic value of a number of parameters of a model, including conditions of excitation of respiratory sound, and the acoustic parameters of the conducting medium is investigated. The advantages of spectral-correlation methods of classification over purely spectral methods are demonstrated. The stability of qualitative criteria of the classification system to inaccuracies in specifying the parameters of the respiratory tract and errors in determining the location of sound recording points is investigated.