Summary
Objective: In an effort to identify and characterize the environmental factors that affect the
number of patients with acute diarrheal (AD) syndrome, we developed and tested two
regional surveillance models including holiday and weather information in addition
to visitor records, at emergency medical facilities in the Seoul metropolitan area
of Korea.
Methods: With 1,328,686 emergency department visitor records from the National Emergency Department
Information system (NEDIS) and the holiday and weather information, two seasonal ARIMA
models were constructed: (1) The simple model (only with total patient number), (2)
the environmental factor-added model. The stationary R-squared was utilized as an
in-sample model goodness-of-fit statistic for the constructed models, and the cumulative
mean of the Mean Absolute Percentage Error (MAPE) was used to measure post-sample
forecast accuracy over the next 1 month.
Results: The (1,0,1)(0,1,1)7 ARIMA model resulted in an adequate model fit for the daily number of AD patient
visits over 12 months for both cases. Among various features, the total number of
patient visits was selected as a commonly influential independent variable. Additionally,
for the environmental factor-added model, holidays and daily precipitation were selected
as features that statistically significantly affected model fitting. Stationary R-squared
values were changed in a range of 0.651-0.828 (simple), and 0.805-0.844 (environmental
factor-added) with p<0.05. In terms of prediction, the MAPE values changed within
0.090-0.120 and 0.089-0.114, respectively.
Conclusion: The environmental factor-added model yielded better MAPE values. Holiday and weather
information appear to be crucial for the construction of an accurate syndromic surveillance
model for AD, in addition to the visitor and assessment records.
Citation: Kam HJ, Choi S, Cho JP, Min YG, Park RW. Acute diarrheal syndromic surveillance –
effects of weather and holidays. Appl Clin Inf 2010; 1: 79–95 http://dx.doi.org/10.4338/ACI-2009-12-RA-0024
Keywords
Surveillance - diarrhea - forecasting - environment - emergency service hospital
