Autumn Annual Conference of IEIE, 2024 (대한전자공학회 추계학술대회)

환자 메타데이터를 활용한 호흡 및 음성 소리의 대조 학습을 통한 SARS-CoV-2 양음성 및 중증도 진단

Diagnosis of SARS-CoV-2 Positivity and Severity Using Contrastive Learning on Respiratory and Voice Data with Patient Metadata

Seoyeon Oh¹ , Dayoung Kim² , and Yelin Kim ^3†*
1Seoul Women's University  ²Ewha Womans University  ³Hongik University 
[Paper] [Code] [Notion]

당신의 목소리는 코로나를 알고 있다!
Contrastive Learning으로 호흡음과 음성 소리, 환자의 메타데이터를 활용해 COVID-19의 양음성 진단과 중증도 진단하기

Proposed Model Architecture

🔮 Abstract

• This study aims to develop a COVID-19 (SARS-CoV-2) diagnosis model using a contrastive learning based on patients' respiratory and voice data.
• Apply a contrastive learning techniques to respiratory and voice data by incorporating patient metadata
  • such as gender, symptoms, and respiratory disease history.
• Not only predicts COVID-19 positivity/negativity but also assesses the severity of the disease.
• Experimental results indicated that incorporating COVID-19-related metadata significantly enhanced diagnostic accuracy.
  • In particular, a history of respiratory disease proved to be a critical factor in predicting severity.

📝 Setting

# Clone the repository
git clone https://github.com/sohds/covid19-diagnosis-using-cough-vowel.git
cd covid19-diagnosis-using-cough-vowel

# Install the dependencies
# For Run Streamlit Code
pip install -r requirements.txt

# Streamlit Code
streamlit run streamlit/app_local.py

📁 Dataset

• Coswara: A respiratory sounds and symptoms dataset for remote screening of SARS-CoV-2 infection

MFCC of vowel 'O'

📝 Results

🗞️ Poster

📚 References

[1] 보건복지부, "비대면진료 시범사업 지침 개정안," April 2024.
[2] Faustino, P, et al, "Crackle and Wheeze Detection in Lung Sound Signals Using Convolutional Neural Networks," Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), 43rd, pp. 345–348. 2021.
[3] Tong, j. y, Sataloff, r. t, "Respiratory Function and Voice: The Role for Airflow Measures," Journal of Voice, Vol. 36, no. 4, pp. 542–553, 2022.
[4] 김철용, "[해외뉴스] 日 스마트폰으로 환자 호흡음 취득 의료기기," 한국의약통신, Link, April 2022.
[5] 장세민, "에이아이포펫, 국내 첫 AI 기반 수의사 비대면 진료 서비스 론칭," AI타임스, Link, March 2024.
[6] Aytekin. I, et al, "Covid-19 Detection from Respiratory Sounds with Hierarchical Spectrogram Transformers," IEEE Journal of Biomedical and Health Informatics, Vol. 28, no. 3, pp. 1273–1284, 2023.
[7] Despotovic v, et al, "Detection of COVID-19 from Voice, Cough and Breathing Patterns: Dataset and Preliminary Results," Computers in Biology and Medicine, 2021.
[8] 대한중환자의학회대한결핵 및 호흡기학회대한감염학회대한항균요법학회, "중증 코로나19 감염(COVID-19) 환자 진료 권고안", Vol. 1, 2021.
[9] Bhattacharya. d, et al. "Coswara: A Respiratory Sounds and Symptoms Dataset for Remote Screening of SARS-CoV-2 Infection. Computers in Biology and Medicine," 2023.