High-Performance Computing in Pandemic Modeling and Simulations

Abstract

High-performance computing (HPC) has become central to modeling pandemics at population scale with high fidelity and responsiveness. This review synthesizes major pre-2019 contributions, from GPU-accelerated network simulations to interactive epidemic modeling frameworks. Key studies include agent-based contagion simulations using EpiSimdemics with GPU offload, demonstrating application speedups of 3.3× on single nodes and up to ~11× across clusters (with latency hiding)PMCSAGE Journals. Interactive platforms like Indemics enable real-time policy intervention modeling through web interfaces, with minimal performance overheadPubMedPMC. Populationscale forecasting models like GLEAM, running thousands of realizations, require HPC for timely simulations—on 20-core clusters, 2,000 runs take 3–5 hoursEurope PMC. The HPC-driven modeling workflow typically includes data ingestion, model calibration, parallel simulation, and visualization. While benefits such as scalability, rapid execution, and high-resolution outputs are clear, challenges persist—resource demands, code complexity, and communication bottlenecks limit broader usability. We conclude HPC is indispensable for modern pandemic simulations. Future efforts should explore cloud–HPC hybrid architectures, GPU-centric modeling, dynamic simulation adjustment, and streamlined pipelines for real-time policy support.