Time Series Analysis and Forecast of Influenza Cases for Different Age Groups in Phitsanulok Province, Northern Thailand

Influenza is still a major problem in Thailand where the incidence varies among age groups. This descriptive research using retrospective data collection aimed to describe the distribution and synchrony, and to forecast Influenza cases in different age groups in Phitsanulok province, northern Thailand. Influenza cases from January 2009 to December 2016 were obtained from R506, Bureau of Epidemiology. Temporal distribution was visually interpreted from line and decomposed graphs. The synchrony between all pairs of age groups was analyzed using Pearson correlation. The 2017 Influenza cases were forecasted using the seasonal ARIMA model in RStudio version 1.1.419. The results showed that trend of Influenza cases for the three age groups: less than 25 years, 25-64 years, and 65 years and older, slightly decreased from 2011 to 2015 and dramatically increased in 2016. The two peaks were observed, i.e. major peak in September and minor peak in February. The cyclic pattern likely observed major peak in two consecutive years for every five years. All pairs of data series co-varied over time. The best models to forecast Influenza cases were seasonal ARIMA (1,0,1)(0,1,1)¹², seasonal ARIMA (1,0,0)(0,1,1)¹² and seasonal ARIMA (1,0,1)(1,1,1)¹² for age less than 25 years, age 25-64 years and age 65 years and older with the MAPE 15.54, 17. 27 and 15.61 respectively. There were 1,698 forecasted cases in age less than 25 years, followed by 1,478 cases in age 65 years and older and 471 cases in age 25-64 years.  The major peak in February and minor peak in September were observed in all age groups. In 2017, the forecasted cases were lower than the reported cases in all data series, except for age 65 years and older.

Keywords: Influenza; time series; forecast; seasonal ARIMA; surveillance

*Corresponding author: Tel.: +66 23 54 85 43 Fax: +66 23 54 85 43 ext. 4777

             E-mail: tsilawan@gmail.com