Monte Carlo, design of experiment, and neural network modeling of basic reproduction number in disease spreading system

Abstract

© 2018 IEEE. In this work, the disease spreading behavior as well as the basic reproduction number were investigated using susceptible-infected-recovered (SIR) model. The disease transmission activity was simulated using Monte Carlo simulation and analyzed using design of experiment and Neural Network. The investigated systems were considered as discrete cells for allocating the agents (population of the system). Each agent was allowed to wander around in carrying out disease transmission. The system sizes and the population (agent) densities were varied to observe the finite size effect, while the infectious period was varied to observe its influence on disease transmission dynamics. Number of agents in SIR states, and number of new infected cases caused by the first infected agent (basic reproduction number) were recorded. From the results, the number of agents in each state as a function of time was found to depend on all considered parameters. Specifically, the main effect plot suggests the basic reproduction maintains with the increased system size, somewhat increases with increasing the density, and mainly increases (at the beginning) with increasing the infectious period. The Neural Network was then used to establish relationship among parameters, where optimized network architecture was found at 3-28-9-1. The accuracy of the network was confirmed via R2and mean absolute value. With Neural Network predicted data, the pair-relationship of inputs to the output was elaborated via interaction plot, giving more insight into the disease spreading characteristic. The residual plot analysis was also performed to confirm the quality of data prediction obtained. With high level of accuracy obtained for predicting data, the results then imply the validity of using multiple modeling/analysis techniques, i.e. Monte Carlo, design of experiment and Neural Network, as supplemental essential tools to model the dynamics of SIR disease spreading.