Application of Horizontal Visibility Graph in Time Series Analysis of Emergency Department Diseases and Mining of Infectious Disease Characteristics

Abstract

The conversion of time series into visualized networks is one of the most important tools for comprehending data patterns and trends. This study pioneers the application of a Horizontal Visibility Graph (HVG) algorithm to transform hospital emergency department time series into complex networks. By analyzing the topological characteristics of networks across different disease categories, we found that all networks exhibit significant small-world properties. Moreover, we observed that the average degree of the networks is notably higher for respiratory diseases. Most importantly, networks of respiratory diseases demonstrate larger maximum eigenvalues of the Laplace matrix, which are closely associated with their stronger infectious potential. These findings reveal the structural signatures of disease spread and provide a critical analytical tool for building network-based early warning systems for specific infectious diseases and triage optimization.