Respiratory Infection Detection Using Cough Pattern Analysis

View at publisher Summarise with VIRAT
Authors: K. Prithivi, S. Gomathy, K. Nithya Pooja, M. Mohamed Hashim, B. Santhosh
Cited by: 0
References: 8

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

The Early Respiratory Disease Detection System employs voice analysis via deep learning to provide a novel, efficient and accessible way to diagnose respiratory disorders that is non-invasive. To initiate the system, a comprehensive audio dataset is compiled to reflect a variety of respiratory diseases, ages, and demographics. Using the Librosa python library, the dataset is analyzed to extract key acoustic features including Mel-Frequency Cepstral Coefficients (MFCCs), spectral contrast and chroma features indicating unique patterns of breath and cough sound. Each audio sample is coupled with a label indicating the corresponding respiratory condition to enable supervised learning. A Convolutional Neural Network (CNN) is subsequently established and trained on the extracted feature set to complicate patterns between sound characteristics and respiratory patterns. The trained model can either classify in real time or batch and can isolate healthy states from diseased respiratory states. An independent test dataset is employed to assess the accuracy of the model and to guide future enhancement for improved reliability and generalization. By taking advantage of CNNs feature-learning advantages, this approach provides a scalable, low cost and reliable diagnostic solution in low-resource or remote healthcare settings. Continuous monitoring, retraining and clinical validation promote consistency and flexibility in a variety of populations and respiratory conditions over time.

Fields of Study & Subjects

Phonocardiography and Auscultation Techniques Artificial Intelligence in Healthcare Music and Audio Processing

Publication details

Journal / source
International Conference on Communication, Computing and Electronics Systems (ICCCES)
Pages
1658–1664
Publication date
21 January 2026
Language
EN