Deep Learning Techniques for Detection of COPD

Deep COPD, an innovative deep learning approach for accurate detection of Chronic Obstructive Pulmonary Disease (COPD) using respiratory sound analysis. The proposed approach utilizes a Convolutional Neural Network (CNN) model trained on a respiratory sound database containing wheezes, crackles, and both crackles and wheezes. To overcome the challenge of a small dataset, innovative techniques such as device-specific fine-tuning, concatenation-based augmentation, blank region clipping, and smart padding are employed. These techniques enable efficient utilization of the dataset, resulting in an impressive accuracy of 90% to 95%.