Automated breath sound analysis

A new approach to analyzing breath sounds is presented. Detailed spectral information of breath sounds was used to investigate the possible relation between lung sound and airways obstruction. Tracheal sound recordings were obtained from two subjects during three sessions of allergen induced airways obstruction. The results of I-nearest neighbour classification clearly demonstrated a relation between lung sound and lung function (FEV₁), within the individual subject. Subjects were breathing normally and no air-flow information was used. The possibility of sound observation during normal breathing would provide a valuable diagnostic tool. From these preliminary results it may be carefully stated that “automated stethoscopy”, tracking the onset of airways obstruction by means of computer aided sound analysis, has become more feasible