多元正態空間掃描統計量模型在探測地方病最強聚集性中的應用

研究了在醫療資源有限的條件下,如何在多組人群中探測某疾病最強地理聚集性的問題.為了有效地探測疾病爆發,只需特別監測和評估某些特定人群即可,這類人群稱為具有最強聚集性的人群（most severe cluster,MSC）.由于各組人群會相互影響,因而提出了一種多元正態空間掃描統計量（multivariate normal scan statistic,MNSS）模型,并以美國紐約州肺癌患者為例驗證了該模型的適用性. This paper focuses on the problem of detecting ageographical cluster with the most severe status in multiple groups of populations with the consideration of limited medical resources. In an early stage of a disease, an outbreak may only be present in some specific population group. Therefore, to efficiently detect the outbreak, specific groups need to be particularly monitored and evaluated. The objective of detection as the most severe cluster (MSC) is defined. Considering interaction among population groups, a multivariate normal scan statistic is proposed. The method is applied to an example of lung cancer in New York State to detect the MSC with a high mortality rate at the aggregate level.