Klasifikasi Penyakit Pernapasan berdasarkan Suara Paru-Paru menggunakan Probabilistic Neural Network

Abstrak

This study aims to classify respiratory diseases using lung sound data analyzed with Probabilistic Neural Network (PNN) method. Lung sound recordings were processed through pre-processing, feature extraction using Mel-Frequency Cepstral Coefficients (MFCC), data resampling, and classification. The research utilized secondary data from the "Respiratory Sound Database" with 920 samples of lung sounds categorized into various respiratory diseases such as asthma, bronchiolitis, and pneumonia. The model achieved a highest accuracy of 98.88%, with an average accuracy of 98.51%, an average precision of 98.54%, and an average recall of 98.51%. These results demonstrate that the PNN method is quite effective in identifying respiratory diseases, offering a potential diagnostic aid in healthcare.