網路導向式計算流行病學：整合疾病動態與人類社會網路的多層次傳染病模型架構

Abstract

網路導向式計算流行病學利用電腦與理論或真實網路拓樸結構研究人類疾病動態和社會趨勢。本論文的主旨在於探討網路導向式計算流行病的重要性、研究現況、優勢與建模過程，並詳述三項原創研究。首先，第一項研究以理論探討無尺度網路下個體資源和疾病傳播成本對於疾病傳播關鍵門檻值的影響，並於流行病模型的基礎上提出解析方程式來解釋關鍵門檻值在無尺度網路下的存在性。該研究指出個體資源和疾病傳播成本的控管對於在無尺度網路下疫情擴散防治的可行性。其次，第二項研究提出整合真實社會網路、個體觀點、國家觀點的多層流行病學架構－多層流病動態模擬器（MEDSim），並以該架構模擬2009年A型H1N1流感疫情在台灣爆發的情形，測試該架構對於不同爆發地點和傳染阻絕方案的靈活性，希望藉此釐清複雜的個體接觸行為對於疾病傳播動態的影響。最後，在第三項研究中分析網路導向式計算流行病學的潛在優勢，並針對網路導向式計算流行病學初學者給予建立網路導向式流行病模型的方法。該研究的目標在於協助擁有較低電腦技能者建立流行病學模型、決定合適的模擬參數與建立操作流程。本論文期望透過上述三項研究，利用電腦模擬來分析多層次的個體互動行為，進而協助傳染阻絕政策的制定。

Network-based computational epidemiologists use computers and either theoretical or actual network topologies to study the transmission dynamics of human diseases and social trends. In this dissertation I discuss the importance, current status, advantages, and modeling procedures of network-based computational epidemiology, specifically presenting three original studies in detail. The first study is an investigation of how resources and transmission costs influence diffusion dynamics and tipping points in scale-free networks. An epidemic model based on an analytic equation is proposed to explain the existence of epidemic critical thresholds in scale-free networks. Study results suggest the possibility of controlling the spread of epidemics in scale-free networks by manipulating resources and costs associated with an infection event. In the second study, a proposal for a multilayer epidemiological framework that integrates realistic social networks, called the Multilayer Epidemic Dynamics Simulator (MEDSim), is described from individual and national perspectives. Model flexibility and generalizability are tested using outbreak locations and intervention scenarios for the 2009 A/H1N1 influenza epidemic in Taiwan. The results coincide with the dynamic processes of epidemics under different intervention scenarios, thus clarifying the effects of complex contact structures on disease transmission dynamics. In the third study, the potential benefits of epidemic simulations and instructions for building network-based epidemic models by novices learning network-based computational epidemiology approaches is investigated. The goal is to help individuals with less advanced computing skills build epidemiological models, determine appropriate simulation parameters, and construct operational procedures. It is my hope that the studies presented in this dissertation can assist in efforts by public health organizations to correctly implement intervention strategies by using simulations to analyze multilayer interactions.