Symphony of Health: AI Unveiling Lung Sound Secrets

Respiratory disease has been a major contributor to deaths worldwide. Early prediction would help to improve the survivability of the patients. This paper deals with the evaluation of lung sounds as a primary step before determining the appropriate course of treatment. An integrated model is employed to detect various respiratory sounds. A Mel-frequency Cepstral Coefficients (MFCC) and 2D convolution neural network is employed, in this work, to identify the lung sounds at early stages of anomaly. Machine Learning algorithms such as Random Forest and Support Vector Machine have been engaged to procure accurate classification of Lung sounds namely crackles, wheezes. The SVM model reported an impressive accuracy of 97.2% followed by Random Forest with accuracy of 96.7%. This innovative approach has the potential for early lung sound detection and classification, making it as a valuable tool for health care professionals.