Scholarlink

The COVID-19 pandemic has resulted in over 520 million confirmed cases and more than 6 million deaths as of May 2022. Despite advancements in vaccine development, there is still a critical need for effective and accessible diagnostic methods. One promising area of research is analyzing cough sounds, which has shown a strong correlation with COVID-19. This study aims to offer insights to help researchers and practitioners choose the most suitable machine learning models, enhancing their effectiveness and efficiency across various fields. This study contributes to the development of more effective applications by guiding model selection. This involves pre-processing audio data, extracting pertinent features, and training various models. Using datasets such as COUGHVID and COVID-19 Recordings, the research highlights the challenges posed by dataset imbalances on model accuracy, as only 72–76% of accuracy in large dataset with some models achieving only 2–3% of F1 Score in COVID label, and 88–96% of accuracy in small dataset with 62% in F1-Score of COVID label. Despite these challenges, the evaluation provides valuable insights into the strengths and limitations of different machine learning models in the context of COVID-19 detection. This study provides a comparative analysis of machine learning models for COVID-19 detection using cough sounds, presenting crucial outcomes that guide the selection of effective AI-driven diagnostic tools for healthcare applications.

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

Evaluating Ensemble Learning and Neural Network Models Towards Covid-19 Detection Through Cough Recordings

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • William Axel Cuangdinata,
  • Kent Alber Fredson,
  • Alfi Yusrotis Zakiyyah,
  • Meiliana

摘要

The COVID-19 pandemic has resulted in over 520 million confirmed cases and more than 6 million deaths as of May 2022. Despite advancements in vaccine development, there is still a critical need for effective and accessible diagnostic methods. One promising area of research is analyzing cough sounds, which has shown a strong correlation with COVID-19. This study aims to offer insights to help researchers and practitioners choose the most suitable machine learning models, enhancing their effectiveness and efficiency across various fields. This study contributes to the development of more effective applications by guiding model selection. This involves pre-processing audio data, extracting pertinent features, and training various models. Using datasets such as COUGHVID and COVID-19 Recordings, the research highlights the challenges posed by dataset imbalances on model accuracy, as only 72–76% of accuracy in large dataset with some models achieving only 2–3% of F1 Score in COVID label, and 88–96% of accuracy in small dataset with 62% in F1-Score of COVID label. Despite these challenges, the evaluation provides valuable insights into the strengths and limitations of different machine learning models in the context of COVID-19 detection. This study provides a comparative analysis of machine learning models for COVID-19 detection using cough sounds, presenting crucial outcomes that guide the selection of effective AI-driven diagnostic tools for healthcare applications.