Deteksi Suara Paru-Paru Menggunakan MFCC (Mel Frequency Cepstral Coefficient) dan M-KNN (Modified K-Nearest Neighbor)

Abstract

The lungs are one of the most important organs in humans because they can meet the body's need for oxygen. It is estimated that hundreds of thousands to millions of the world's population are affected by lung disease every year. According to WHO, lung disease is one of the top 10 causes of public health problems in the world. One way that doctors use to diagnose lung disorders is by listening to the sound of breathing in the lungs using a stethoscope with acculturation techniques. Accurate recognition of lung conditions is needed so that it becomes a basic screening of people who can have abnormalities or not in lung conditions. This study focuses on researching and trying the Modified K-NN method in determining the lung condition of a person. The steps taken in this research are preprocessing, feature extraction and matching with the Modified K-NN algorithm. Then testing with 5 fold validation and confusion matrix, so that the largest results are obtained in fold 4, namely 0.98 or 98% with precision 1, recall 0.14, f1-score 0.25 for normal and 0.97 precision, recall 1, f1-score 0.98 for abnormal 97.83 %