Keywords: Data correlation, Disease surveillance, Early detection, Nonparametric methods, Sequential monitoring, Spatio-temporal data
Online sequential monitoring of the incidence rates of chronic or infectious diseases is critically important for public health and stability of our society. Governments around the world have invested a great amount of resources in building global, national and regional disease reporting and surveillance systems. In these systems, conventional control charts, such as the cumulative sum (CUSUM) and exponentially weighted moving average (EWMA) charts, are usually included for disease surveillance purposes. However, these charts require many assumptions on the observed data, including the ones that the observed data are independent and follow some parametric distributions when no disease outbreaks are present. These assumptions are hardly valid in practice, making the results from the conventional control charts unreliable. Motivated by an application to monitor the Florida influenza-like illness data, we develop a new sequential monitoring approach in this paper, which can accommodate the dynamic nature of the disease incidence rates (i.e., the disease incidence rates change over time due to seasonality and other reasons), spatio-temporal data correlation, and arbitrary data distribution. It is shown that the new method is much more reliable to use in practice than the commonly used conventional charts for sequential monitoring of disease incidence rates. The proposed method should be useful for many other applications, such as spatio-temporal monitoring of air quality in a region and monitoring of sea-level pressure in oceanography.