A bitstring approach for implementing agent-based epidemiological models

Abstract

Agent-based Models (ABM) are gaining importance over traditional epidemiological modeling due to advances in computing technology and by the need for detailed epidemiological analysis of emergent diseases. Unfortunately, the advantages of ABMs are achieved at the cost of significantly large execution times and high memory consumption for large-scale simulations. Addressing the memory issue, we designed and implemented an ABM using an innovative feature: the bitstring approach. In this approach, the attributes of the agents are represented by an array of bits instead of using traditional data structures. We describe the bitstring data representation and present a suitable logical formulation to map conceptual and compartmental models to a computer implementation by using spatio-temporal operators that represent the agents behavior and the disease propagation. Versions for CPU and GPU were implemented and presented good qualitative results and behavior similar to those obtained by traditional versions. The application of the bitstring technique proved to be relevant in economy of memory, allowing to store the same attributes using up to 80% less memory space. Besides, the use of the proposed approach also improved the data copy time between CPU-GPU in the GPU implementation, reducing the execution time up to 20%.