Federated learning based Covid-19 detection

Area of Science:

• Artificial Intelligence
• Medical Imaging
• Public Health

Background:

• The global COVID-19 pandemic necessitates accurate and accessible diagnostic tools.
• Existing diagnostic methods may face challenges in speed and data privacy.
• The integration of deep learning and federated learning offers potential for secure and efficient medical diagnostics.

Purpose of the Study:

• To develop a deep and federated learning model for rapid COVID-19 detection from chest X-ray images.
• To prioritize user data security while maximizing diagnostic accuracy.
• To leverage the Internet of Medical Things (IoMT) for improved healthcare access.

Main Methods:

• A sequential Convolutional Neural Network (CNN) model was developed using deep and federated learning principles.
• The architecture was designed for efficient client and server-side operations, utilizing StreamLit for the front-end and Flower for the back-end.
• The model was trained over three federated communication rounds.

Main Results:

• The proposed model achieved a global accuracy of 99.59% in detecting COVID-19 from chest X-ray images.
• The federated learning approach ensured user data security during the training process.
• The system provides rapid COVID-19 detection within seconds.

Conclusions:

• The developed deep and federated learning model offers a highly accurate and secure solution for COVID-19 detection.
• The integration with IoMT enhances the accessibility of rapid diagnostic services.
• This approach holds significant potential for improving healthcare delivery during pandemics and beyond.