請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/91697
標題: | 在早期流行病傳播過程中自相似性之貝氏樹演算法 A Bayesian Tree Algorithm of Self-similarity in Early Stage Epidemic Transmission Process |
作者: | 王成驊 Cheng-Hua Wang |
指導教授: | 溫在弘 Tzai-Hung Wen |
關鍵字: | 冪次律,早期階段,傳播樹,自相似,貝氏推論,增階演算法 (OIA), power-law,early stage,transmission tree,self-similar,Bayesian inference,Order-increasing Algorithm (OIA), |
出版年 : | 2024 |
學位: | 碩士 |
摘要: | 在流行病早期階段的冪次律成長現象描述了經驗的規模關係,隱含傳播過程潛在著自相似機制。除此之外,相較於傳統的隔室模型,由傳播鏈與感染者組成的傳播樹能夠提供研究人員更多流行病的圖像化資訊。由於現有研究缺乏使用傳播樹揭示流行病中的自相似性,本研究目的為尋找在流行病早期的傳播樹自相似模式。本研究發展增階演算法 (OIA)推論條件式感染連結度集合來作為一個自相似模式。OIA已被施行於48項因子水準的模擬研究以尋找自相似模式、和一個真實流行病案例以重建流行病的傳播樹。結果發現末端節點的增加會打破早期流行病真實傳播樹中自相似特徵,並減低重建過程的品質。除此之外,OIA在尋找模式與重建過程中很穩健,其意涵為辨識出來的自相似模式能夠用於比較不同流行病在早期階段的傳播特性比如超級傳播者及其出現的可能性。故OIA作為一個貝氏樹演算法,成功地在模擬研究與流行病真實案例中推論出條件式感染連結度之重複模式以達成本研究目的。 A power-law phenomenon of growth in the early stage of epidemics describes an empirical scaling relationship, suggesting a self-similar mechanism underlying the spreading process. In addition, the transmission trees formulated by transmission chains and infectives can provide researchers with more graphical information on epidemics than conventional compartment models. Due to current studies lacking the use of transmission trees to unveil the self-similarity in epidemics, the purpose of this article is to identify self-similar patterns of transmission trees in the early stage of epidemics. This article developed an Order-increasing Algorithm (OIA) inferring the sets of conditional infectious connectivity as a self-similar pattern. The OIA was implemented in both simulation studies of 48 levels of factors to identify self-similar patterns and a real epidemic case to reconstruct transmission trees for epidemics. The finding highlighted that the increasing of leaf nodes would break the self-similar feature in real epidemic transmission trees in the early stage and reduce the quality of the reconstruction process. In addition, the OIA was robust in the processes of identification and reconstruction, implying that the identified self-similar patterns could be used to compare spreading characteristics, such as super-spreaders and their likelihood of occurrences, among different epidemics in the early stage. In conclusion, the OIA, as a Bayesian tree algorithm, successfully infers the repeated patterns of conditional infectious connectivity in both simulation studies and a real epidemic case to fulfill the purpose of this article. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/91697 |
DOI: | 10.6342/NTU202400399 |
全文授權: | 同意授權(全球公開) |
顯示於系所單位: | 統計碩士學位學程 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-112-1.pdf | 3.35 MB | Adobe PDF | 檢視/開啟 |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。