A HYBRID CNN–LSTM APPROACH FOR EFFICIENT RESPIRATORY DISEASE CLASSIFICATION USING GTCC AND STFT FEATURES

Respiratory diseases remain a major cause of global mortality and morbidity, placing a heavy burden on healthcare systems.Early and accurate detection is vital to improve patient outcomes and reduce death rates.This study proposes a hybrid deep learning framework that integrates Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks for classifying respiratory conditions from lung sound recordings.The approach incorporates Gammatone Cepstral Coefficients (GTCC) and Short-Time Fourier Transform (STFT) features to capture complementary spectral and temporal information.The CNN component extracts spatial patterns from spectrogram representations, while the LSTM captures temporal dependencies across audio segments.The model was trained and validated on a publicly available lung sound dataset after applying preprocessing techniques such as noise reduction and feature normalization.Performance was assessed using accuracy, precision, recall, and F1-score, with results showing the proposed CNN-LSTM architecture outperforming baseline CNN-or LSTM-only models.These findings demonstrate the potential of deep learning-based auscultation analysis as a cost-effective, non-invasive, and efficient screening method for respiratory diseases, particularly valuable in resource-limited healthcare environments.