Category
Applied
Description
The aim of this research is to utilize a biostatistical methodology for the purposes of identifying influenza trends during an early stage in the U.S. utilizing national surveillance data collected by the Centers for Disease Control and Prevention's Influenza-Like Illness Surveillance Network (ILINet) for the years 2010-2024. Descriptive statistics were used to analyze weekly influenza-like illness (ILI) activity, along with other methodologies including: moving averages to smooth out fluctuations in ILI activity; autoregression models to evaluate temporal dependence in ILI activity; and threshold detection methods to determine when ILI activity reached epidemic levels. The results show that there are strong seasonal patterns and significant temporal dependence associated with influenza activity. Also, through the utilization of an early signal detection method that utilizes a 3 week moving average, higher sensitivity was found and that ILI activity at epidemic levels occurred significantly sooner than what would have been identified with the use of traditional threshold detection methods. Overall, these results indicate that transparent and replicable statistical techniques can be valuable for improving situational awareness and for developing improved public health surveillance systems. Keywords: Influenza; Influenza-Like Illness (ILI); Epidemiological Surveillance; Biostatistics; Time-Series Analysis; Autoregressive Modeling; Early Outbreak Detection; CDC ILINet.
Biostatistical Approaches for Early Identification of Influenza Trends in the United States Using CDC Surveillance Data
Applied
The aim of this research is to utilize a biostatistical methodology for the purposes of identifying influenza trends during an early stage in the U.S. utilizing national surveillance data collected by the Centers for Disease Control and Prevention's Influenza-Like Illness Surveillance Network (ILINet) for the years 2010-2024. Descriptive statistics were used to analyze weekly influenza-like illness (ILI) activity, along with other methodologies including: moving averages to smooth out fluctuations in ILI activity; autoregression models to evaluate temporal dependence in ILI activity; and threshold detection methods to determine when ILI activity reached epidemic levels. The results show that there are strong seasonal patterns and significant temporal dependence associated with influenza activity. Also, through the utilization of an early signal detection method that utilizes a 3 week moving average, higher sensitivity was found and that ILI activity at epidemic levels occurred significantly sooner than what would have been identified with the use of traditional threshold detection methods. Overall, these results indicate that transparent and replicable statistical techniques can be valuable for improving situational awareness and for developing improved public health surveillance systems. Keywords: Influenza; Influenza-Like Illness (ILI); Epidemiological Surveillance; Biostatistics; Time-Series Analysis; Autoregressive Modeling; Early Outbreak Detection; CDC ILINet.
