A COVID-19 Auxiliary Diagnosis Based on Federated Learning and Blockchain

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Data Security in Medicine

Background:

• The high transmission and pathogenicity of COVID-19 necessitate rapid and accurate diagnostic methods.
• Existing diagnostic approaches face challenges related to data sharing, privacy, and security among medical institutions.
• Data silos and insufficient medical data hinder the development of robust diagnostic models.

Purpose of the Study:

• To propose an auxiliary diagnostic method for COVID-19 utilizing federated learning and blockchain.
• To enable effective collaborative model training among multiple medical institutions while addressing data sharing difficulties.
• To enhance data privacy, security, and data rights for participating medical institutions.

Main Methods:

• Application of federated learning for COVID-19 medical diagnosis to refine big data value and overcome data silos.
• Integration of blockchain technology to protect sensitive information and ensure data rights, mitigating third-party dependence.
• Simulation of realistic scenarios using a real COVID-19 dataset to analyze model validity, applicability, and iteration delays.

Main Results:

• Demonstration of multiparty participation in model training.
• Validation of enhanced data protection through the proposed method.
• Experimental evidence supporting the effectiveness of the combined federated learning and blockchain approach for COVID-19 diagnosis.

Conclusions:

• The proposed federated learning and blockchain-based method effectively facilitates collaborative COVID-19 diagnosis.
• This approach successfully addresses challenges in data sharing, privacy, and security in medical big data.
• The method aids medical personnel in diagnosing coronavirus disease more effectively, improving outbreak response.