Keywords
MFCC, Artificial Intelligence, Forced Vital Capacity, Expiration
Abstract
Objective: In this study, it is aimed to calculate the tidal volume capacity with mobile infrastructures and to determine the O2 and CO2 values from breath, phonation and cough sounds with the help of artificial neural networks using voice feature vectors. In addition, it is aimed to develop a method that will enable computation of forced expiratory volume via mobile devices.
Materials and Methods: Since sound disease diagnosis is basically a classification problem, tidal volume classification is considered as a pattern recognition problem. Our automatic O2 and CO2 classification study as a pattern recognition function includes three subcomponents; Identifying and obtaining features, Feature selection and Classifiers.
Results: In the study, a new dataset created from two different datasets was used; MFCC (Mel Frequency Cepstral Coefficients) Mel frequency kepstral coefficients of each sound in this new dataset were extracted and converted into png format spectrogram graphs. Sounds converted to picture format are classified by Convolutional Neural Networks (CNN), one of the deep learning algorithms in artificial intelligence. As a result, a very high performance has been achieved with the CNN network.
Conclusion: In this study, a system has been developed for the detection of voice-based lung capacity, O2 taken into the vital area and CO2 values exhaled by using artificial intelligence. Each image-converted sound is analyzed using Convolutional Neural Network (CNN), one of the algorithms in deep learning, which is one of the sub-branches of feature extraction and artificial intelligence. For the selection of this algorithm, studies with similar usage areas were taken into consideration; As a result of the inferences obtained from these studies, a path was followed. At the end of the system developed with the CNN network, a success of 99.60% was achieved. This performance shows that the created system works with high success.
DOI
10.33107/ijbte.2023.1
First Page
1
Last Page
14
Recommended Citation
Hajrizi, Edmond
(2023)
"Detection of Breath, Phonation and Cough Sounds Using Sound Feature Vectors and Artificial Neural Networks by Calculating Lung Forced Vital Capacity with Mobile Infrastructures,"
International Journal of Business and Technology: Vol. 11:
Iss.
3, Article 1.
DOI: 10.33107/ijbte.2023.1
Available at:
https://knowledgecenter.ubt-uni.net/ijbte/vol11/iss3/1