An Enhanced Deep Learning Approach for Disease Classification from Respiratory Sound

Technological progress has revolutionized the automated classification of respiratory diseases, harnessing the complex acoustic data inherent in the human respiratory system. This data serves as a crucial resource for identifying various respiratory conditions. Traditional diagnostic methods face challenges like high costs and limited awareness, underscoring the need for practical, non-invasive alternatives. Hence, the analysis of respiratory sounds emerges as a cost-efficient solution for effective disease classification in diverse healthcare settings. This paper introduces an optimized deep-learning model tailored explicitly for respiratory disease classification. By utilizing COPD audio data from Kaggle and extracting features such as MFCC and spectrograms, our neural network model achieves outstanding performance, boasting an accuracy of 0.93. This advancement facilitates early detection and timely intervention and addresses the limitations of traditional diagnostic approaches, paving the way for more accessible and accurate healthcare solutions in respiratory medicine.