Detection of COVID-19 Disease Using Federated Learning

The study investigates the use of Federated Learning (FL) in combination with the AlexNet architecture for the early confirmation of COVID-19 from chest radiography data. A thorough literature review highlights the potential of FL in healthcare and its application in COVID-19 detection, emphasizing privacy preservation and data security. The methodology involves collecting and 80% preprocessing a diverse dataset, partitioning the data among multiple clients, and iteratively training a global model. Experimental results demonstrate the effectiveness of the FL-based AlexNet model, achieving a high accuracy of 92% on a test dataset. Future research should focus on expanding datasets, exploring advanced privacy techniques, optimizing and scaling FL models, developing real-time applications, and integrating FL models with other modalities for improved COVID-19 detection. These advancements can contribute to early detection and diagnosis, supporting healthcare professionals in combating the pandemic.