PandemicShield: A Unified AI-Blockchain Framework for Disease Detection, Drug Forecasting, and Traceability

Abstract

This study unveils an innovative system that merges machine learning and Blockchain technologies to tackle essential issues in pandemic scenarios, including the early identification of unknown diseases, prediction of drug requirements, and suggestion of appropriate medications, secure tracking of drug supply chains, and providing clear, interpretable insights. The approach integrates Graph Neural Networks (GNNs) with ensemble learning methods for detecting diseases, employs causal inference alongside ARIMA for forecasting drug needs, utilizes reinforcement learning (RL) for recommending drugs, leverages Ethereum smart contracts with alert mechanisms for supply chain tracking, and applies counterfactual explanations to enhance interpretability. Unlike earlier frameworks that emphasize detection, forecasting, and tracking but overlook actionable insights, this system ensures stakeholders can make informed decisions through explainable outcomes. Utilizing encrypted health records (EHR), MongoDB for data storage, and a React-driven frontend, the system is assessed using a combination of synthetic and real-world data, achieving a detection accuracy of 93.5%, a forecasting error of 3.9%, a 16% improvement in recommendation rewards, and a Blockchain latency of 1.4 seconds. This system delivers a robust, interpretable solution for managing future pandemics effectively