Abstract:
This study aims to generate the equation for predicting the utilization rate of the local
hospital in children aged 1-10 years old with respiratory disease by an artificial neural networks
(ANNs). Using ANNs, the prediction equation was simulated by using retrospectively collected data
including nine environmental and health indicators. The eight indicators were taken from the
existing environmental and health data from 2010 to 2014 of the Air Quality and Sound
Management Office in Thailand report and the Health Promoting Hospital District in Chaloem
PhraKiat District, respectively. The factor related to the activity outside their home was obtained
from the literature. The results showed that only five indicators were the predictors for estimating
the number of children who visited the hospital district with respiratory diseases. These indicators
included season, distance, number of children with respiratory disease, %weight for height and
PM10 levels. The prediction equation had the correlation coefficient about 63.20%, accuracy with
rate of 86.33% and relative error 13.88%. However, the equation was able to be performed for
predicting the low respiratory disease rate, with sensitivity, precision and confidence rates of
92.38%, 89.57% and 77.20%, respectively. It is concluded that the ANNs can be applied to select
the environmental and health indicators that were associated with the respiratory disease; and can
generate the equation from the existing environmental and health datasets to estimate the
number of children with respiratory disease who come to be treated in a hospital district.