Scholarlink

This research aims to fill the current gap of the absence of a fast and non-contact diagnostic approach for COVID-19 by processing signals of coughs. Using enhancements in Machine Learning (ML) and Deep Learning (DL), our study explores an effective model that can diagnose cough sounds related to COVID-19 and disentangle them from other respiratory sounds. The provided dataset incorporates various cough sounds with COVID-positive cases and people with other diseases, so it allows using the proposed algorithm for training and testing. To address the real-time behemoth challenge, the CNNs and RNNs models were combined to come up with a comprehensive model capable of detecting very slight acoustic features that may be an inflection of COVID-19 infection with an accuracy of 98.93%. The study brings value to the design of efficient, inexpensive, and large-scale testing of mass populations at high risk for diseases requiring early intervention. Furthermore, it has possibilities to enhance early diagnosis of COVID-19, as a useful addition to the standard diagnostic procedures, and to help prevent the virus transmission in different public spaces by intervening before contagion.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Cough Signal Processing for Classification and Early Detection of COVID-19 Using Deep Learning Integrated Approach

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Prachi Kaushik,
  • Monica Bhutani,
  • Komal Bhagat

摘要

This research aims to fill the current gap of the absence of a fast and non-contact diagnostic approach for COVID-19 by processing signals of coughs. Using enhancements in Machine Learning (ML) and Deep Learning (DL), our study explores an effective model that can diagnose cough sounds related to COVID-19 and disentangle them from other respiratory sounds. The provided dataset incorporates various cough sounds with COVID-positive cases and people with other diseases, so it allows using the proposed algorithm for training and testing. To address the real-time behemoth challenge, the CNNs and RNNs models were combined to come up with a comprehensive model capable of detecting very slight acoustic features that may be an inflection of COVID-19 infection with an accuracy of 98.93%. The study brings value to the design of efficient, inexpensive, and large-scale testing of mass populations at high risk for diseases requiring early intervention. Furthermore, it has possibilities to enhance early diagnosis of COVID-19, as a useful addition to the standard diagnostic procedures, and to help prevent the virus transmission in different public spaces by intervening before contagion.