Tugas Akhir Elektro
Analisis Suara Pernapasan Paru-paru Asma Dengan Tidak Asma Menggunakan Metode k Nearest Neighbors
Asthma is a disease of the airways that causes an increase in airway hyperesponsiveness and causes
wheeze symptoms. The sound of wheeze breath is one of the characteristics that indicates a person has
asthma. This research was conducted to create and test a system that can identify differences in wheeze
respiratory voice traits in asthma and other respiratory patients with the k-Nearest Neighbors (k-NN) method.
The sound features used are the average signal and the standard deviation in the time domain, the average
spectrum, the standard deviation of spectrum, the highest magnitude at 0Hz frequency, the frequency with
the highest magnitude first, second, and third. k-NN is a method of classifying objects based on learning data
closest to them. Obtained wheeze and non wheeze respiratory voice data through live recording to subjects
with asthma and not asthma. From all the sound data obtained then segmentation of the data to take the
respiratory event needed then carried out the extraction of features to get the mathematical characteristics of
the sound. 80% of the total data was conducted using the 10 fold cross validation method and was reviewed
with maximum classification capabilities at k=3 and k=5 with the same validity of 97.2%. For k-NN
performance testing at the final stage obtained the maximum classification capability for k=3 is 86.6% and
k=5 is 86.6%.
Tidak ada salinan data
Tidak tersedia versi lain