Assessing Seasonal Variation in Multisource Surveillance Data: Annual Harmonic Regression
- Cite this paper as:
- Lofgren E., Fefferman N., Doshi M., Naumova E.N. (2007) Assessing Seasonal Variation in Multisource Surveillance Data: Annual Harmonic Regression. In: Zeng D. et al. (eds) Intelligence and Security Informatics: Biosurveillance. Lecture Notes in Computer Science, vol 4506. Springer, Berlin, Heidelberg
A significant proportion of human diseases, spanning the gamut from viral respiratory disease to arthropod-borne macroparasitic infections of the blood, exhibit distinct and stable seasonal patterns of incidence. Traditional statistical methods for the evaluation of seasonal time-series data emphasize the removal of these seasonal variations to be able to examine non-periodic, and therefore unexpected, or ‘excess’, incidence. Here, the authors present an alternate methodology emphasizing the retention and quantification of exactly these seasonal fluctuations, explicitly examining the changes in severity and timing of the expected seasonal outbreaks over several years. Using a PCRconfirmed Influenza time series as a case study, the authors provide an example of this type of analysis and discuss the potential uses of this method, including the comparison of differing sources of surveillance data. The requirements for statistical and practical validity, and considerations of data collection, reporting and analysis involved in the appropriate applications of the methods proposed are also discussed in detail.
Unable to display preview. Download preview PDF.