Hybrid-based framework for COVID-19 prediction via federated machine learning models.

Area of Science:

• Computer Science
• Artificial Intelligence
• Public Health

Background:

• The COVID-19 pandemic's high transmissibility necessitates advanced detection and monitoring systems.
• Existing systems may lack the real-time processing capabilities required for effective pandemic control.

Purpose of the Study:

• To propose a novel smart COVID-19 disease monitoring and prognosis system.
• To integrate machine learning (ML), cloud, fog, and Internet of Things (IoT) technologies for enhanced pandemic response.

Main Methods:

• Leveraging IoT devices for data collection from medical and non-medical sources.
• Implementing a hybrid fog-cloud framework with distributed batch and stream federated ML as a service (MLaaS).
• Evaluating batch and stream ML algorithms using quantitative (accuracy, precision, RMSE, F1 score) and qualitative (latency, response time) metrics.

Main Results:

• Stream ML algorithms demonstrated potential for integration into COVID-19 prognosis.
• The proposed system enables real-time symptom data processing and prediction within a fog-cloud environment.
• Federated MLaaS models were evaluated for both long-term (cloud) and short-term (fog) decision-making.

Conclusions:

• Stream ML algorithms are suitable for early prediction of suspected COVID-19 cases.
• The integrated fog-cloud framework enhances the efficiency of COVID-19 monitoring and prognosis.
• This approach offers a scalable solution for real-time disease surveillance during pandemics.