運用「台灣通勤人口暨交通運輸網路模型」探討新型流感傳播之最佳交通阻絕策略施行方式

Applying Genetic Algorithms to Evolve The Best Combination of Traffic Prevention Policies Against Pandemic Novel Influenza

陳碩璜(Shuo-Huang Chen)

指導教授 ： 孫春在

國立交通大學/資訊學院/資訊科學與工程研究所/碩士(2009年)

https://doi.org/10.6842/NCTU.2009.01106

摘要

當新型流感於地方爆發感染時，適當的交通阻絕策略能夠延遲傳染病於地域上的擴散並錯開各地區大流行之高峰期，以防止感染人數的爆增，使有限的醫療資源(例如重症病床數)得以控制當下的疫情，讓醫療體系不至於崩潰；此外，疫情的延遲也為政府防疫執行單位爭取時間，規劃佈署其它公衛政策，例如採購抗病毒藥劑、研發疫苗等。
然而，通勤及交通網路結構的異質性會影響新型流感於時間上與地域上的傳播進程，所以必須考量這些要素佈署交通阻絕策略。另一方面，交通阻絕策略能夠帶來防疫成效，卻也造成負面影響，它會阻礙交通、影響當地的經濟活動，而複雜的通勤及交通網路結構使規劃一個造成最小負面影響卻能達到最大成效的交通阻絕策略須要有技術去克服。
本研究提出以基因演算法來優選交通阻絕策略的近似最佳解決方案。並藉由結合通勤及交通網路結構與決定性倉室模型(deterministic compartmental model)，模擬新型流感於台灣爆發感染時的阻絕方法。實驗結果顯示基因演算法能夠依據「新型流感爆發的感染源」與「施行策略的時間點」找到符合效益的交通阻絕策略。除了交通阻絕策略的探討外，本研究的模型與基因演算法也可用來研究其它公衛政策有效的方法，例如數量有限的抗病毒藥劑該如何分配於各個城市，以有效控制疫情。

關鍵字

新型流感 ； 決定性倉室模型 ； 複雜網路結構 ； 基因演算法

並列摘要

An effective traffic prevention policy can delay the spread of the novel influenza and can decrease the synchrony in timing of epidemics among all cities and rural areas so as to prevent a sudden surge of infections. By doing so, the limited medical resources (Such as the wards for critical care) will be adequate to save the medical system from collapsing. Furthermore, the delay can buy some time for the authorities to formulate other intervention policies, including the procurement of anti-viral medicines from pharmaceutical companies and the development of new vaccine. However, the potential costs of a traffic prevention policy must be taken into account. For example, the policy may impedes the traffic and affects local economic activity.
Due to the great complexity of the transportation network, tailoring a cost-efficient traffic prevention policy could be challenging. In this thesis, a genetic algorithm was applied for finding an optimal traffic prevention policy. Because the heterogeneity of the transportation networks affects the temporal and spatial progression of infectious diseases dramatically, we investigate the planning of efficient traffic prevention policies based on the complete topology structure of the transportation network. With the combination of the deterministic compartmental model—which describes local infection dynamics among individuals—and the transportation infrastructure in Taiwan, we are able to stimulate the transmission dynamics of pandemic novel influenza,
Our experimental results show that our proposed genetic algorithm is able to evolve the optimal traffic prevention policies according to the sources of the outbreaks and the timing of applying the policies respectively. In the future, we can employ this model and genetic algorithms to plan other public health policies, such as ways to make proper distribution of finite anti-viral medicines to cities in Taiwan, and thus contain the outbreak of pandemic novel influenza.

並列關鍵字

Novel influenza ； deterministic compartmental model ； Complex Network ； Genetic algorithm

參考文獻

[1] Anderson, R. M., May, R. M., “Infectious diseases of humans: dynamics and control.” Oxford University Press, 1991.

[3] Colizza, V., et al., ”Modeling the Worldwide Spread of Pandemic Influenza: Baseline Case and Containment.” PLoS Medicine, 2007. 4(1): e13.doi:10.1371/journal.pmed.0040013s

[4] C.Y. Huang., et al., “Influences of Resource Limitations and Transmission Costs on Epidemic Simulations and Critical Thresholds in Scale-Free Networks”Simulation, 2009. 85(3): p. 205-219.

[5] Darren, M. Green, Istvan, Z. Kiss, and Rowland, R. Kao, “Parameterization of individual-based models: Comparisons with deterministic mean-field models.”

[6] Ferguson, N. M., et al., “Strategies for containing an emerging influenza pandemic in Southeast Asia.” Nature, 2005. 437: p. 209-214.