APPLICATION OF THE NEURAL NETWORK TECHNOLOGY FOR DETECTION AND MONITORING OF AUSCULTATIVE PHENOMENA IN DIAGNOSIS AND TREATMENT OF DISEASES OF THE RESPIRATORY SYSTEM

Background. Implementation of electronic auscultation in practical medicine seems promising and worthwhile. In the Republic of Belarus this trend is practically not developed. Goal. To study the effectiveness of using the "Lung Passport" neural networks in respiratory diseases diagnostics and on this basis develop an automatic system for assessing the state of the respiratory system. Material and methods. To conduct an electronic auscultation the “Lung Passport” hardware-software system based on the machine learning algorithm for classification of the auscultative phenomenon type was used. Results. The automatic analysis system of sound phenomena has a high sensitivity (80.81%-93.33%) and specificity (83.33%-98.99%) and allows you to objectify auscultative data. Conclusions. The use of the auscultative phenomena automatic classification method based on machine learning will increase the efficiency of early diagnosis and monitoring of the respiratory pathology.