Blockchain enabled collective and combined deep learning framework for COVID19 diagnosis

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Blockchain Technology

Background:

• The rapid spread of SARS-CoV-2 necessitates advanced diagnostic tools for COVID-19.
• Conventional AI diagnostic models face challenges with centralized data, complexity, and privacy concerns, hindering global data sharing.
• There is a critical need for secure, collaborative diagnostic frameworks that balance accuracy with data protection.

Purpose of the Study:

• To introduce a novel framework, the Combined Learning Collective Deep Learning Blockchain Model (CLCD-Block), for secure and accurate COVID-19 diagnosis.
• To address the limitations of centralized data storage and training in AI-driven medical diagnostics.
• To develop a model that enhances data privacy and facilitates secure global data exchange.

Main Methods:

• The CLCD-Block framework integrates blockchain technology with a combined learning paradigm for decentralized AI training.
• A hybrid capsule learning network is employed for accurate diagnostic predictions using lung CT images.
• The model aggregates data from multiple institutions, ensuring secure distribution and reduced complexity.

Main Results:

• The CLCD-Block model demonstrated superior diagnostic performance, achieving an accuracy exceeding 97% on lung CT images.
• On four benchmark datasets, the model achieved high metrics: up to 98.79% Precision, 98.84% Recall, 98.79% Specificity, 98.81% F1-Score, and 98.71% Accuracy.
• The framework effectively balances diagnostic accuracy with robust privacy protection.

Conclusions:

• The CLCD-Block framework offers a secure, decentralized, and accurate solution for COVID-19 diagnosis.
• This model's adaptability extends to other healthcare applications, including chronic disease diagnosis and infectious outbreak monitoring.
• Future research will focus on enhancing scalability and real-time performance for broader healthcare dataset integration.