No abstract

Biosurveillance for rapid detection of epidemics of diseases is a challenging area of endeavor in many respects. Hence, this area is in need of development of methodology and opens to novel methods of detection. In this study, a new simple statistical early outbreak detection approach is proposed to detect outbreaks of diseases in real time. The new approach is called LWMAT since it is based on linearly weighted moving average. Furthermore, it does not require a long baseline and partly takes into account of likely features of biosurveillance data such as nonstationary and overdispersion to some extent. Moreover, this newly proposed method is easily adapted to automated use in public health surveillance systems to monitor simultaneously large number time series of indicators associated with the relevant diseases. To compare the performance of the new method with those of some well‐known outbreak detection methods, semisynthetic data with outbreaks of various magnitudes and durations are simulated by considering the weekly number of outpatient visits for influenza‐like illness for the influenza seasons 2014‐2015 through 2017‐2018 at Centers for Disease Control and Prevention (CDC) in the United States. Under the conditions of the simulation studies, Serfling regression and Farrington flexible seem to be preferable methods for monitoring the weekly influenza data at CDC in terms of early identification of influenza outbreaks with a high probability. In addition, the newly proposed LWMAT‐type methods appear to be promising and useful methods in the case of small magnitude outbreaks with a short duration.

No abstract

No abstract

ÖZETBu çalışmada iki normal dağılımlı rasgele değişken arasındaki korelâsyonun anlamlılık testi için yeni bir test yaklaşımı önerilmektedir. Bu yeni yaklaşım korelâsyonun işareti ile uyumlu olmayan aykırı değerlere karşı sağlamdır. Korelasyonun anlamlılığının testi için kullanılan geleneksel Pearson t-testi ile önerilen test yaklaşımının güç performanslarını karşılaştırmak amacıyla, bu tür aykırı değerlerin birkaç tanesinin varlığı durumunda bir benzetim çalışması gerçekleştirilmiştir. Bu benzetim çalışması sonucunda, önerilen yaklaşımın korelâsyonlar için olan Pearson ttestinden daha iyi performans gösterdiği görülmüştür. Bu tür aykırı değerler örneklemin yaklaşık olarak %5 ya da %6'sından daha fazlasını oluşturmamak kaydıyla yeni yaklaşıma ilişkin bu iyi performansın devam ettiği söylenebilmektedir.Anahtar kelimeler: Kolmogorov-Smirnov uyum iyiliği testi, Cramer-von Mises uyum iyiliği testi, Deneysel güç karşılaştırması, Simülasyon ABSTRACTIn this study, a new testing approach to significance test for correlation between two normally distributed random variables is proposed. This new approach is robust to the outliers that are incompatible with the sign of the correlation. To evaluate and compare power performance of the proposed approach with that of usual the Pearson t-test for correlation in the case of this type of a few outliers, a simulation study is conducted. As a result of the simulation study, it has been found that the proposed approach performs better than the usual t-test in the presence of such outliers. In addition, it seemed that the nice performance of the new approach continues to exist provided that thesde outliers constitute approximately no more than 5% or 6% of the sample.