Predicting Covid-19 Infections with a Multi-Agent Organizational Approach and Machine Learning Techniques

Abstract

Our study presents a strategy for designing and implementing a Multi-Agent System (MAS) using organizational paradigms. The developed system offers a healthcare-oriented approach that utilizes the Internet of Medical Things (IoMT) to assist public health authorities in predicting COVID-19-infected patients. The proposed approach leverages autonomous agents to handle dynamic data from various sources within a structured organization. These agents collaborate to make effective, real-time predictions. As the agents continuously learn from the cases entering the system, the accuracy of predictions improves over time. The system was implemented using the JaCaMo framework, which integrates three key layers of MAS programming: organization, environment, and agent programming. The methodology demonstrated a prediction accuracy of over 90%, outperforming state-of-the-art (SOTA) approaches by enabling faster real-time decision-making. This capability facilitates the efficient processing of real-time big data, making a significant contribution to the advancement of predictive healthcare systems.