Time series forecasting in anxiety disorders of outpatient visits using data mining

Abstract

This study aims to forecast the number of anxiety disorders patients who would be seeking treatment at an outpatient clinic in 2011 by comparing two Artificial Neural Network (ANN) models and selecting the most powerful model. Data were collected from the Prasrimahabhodi Psychiatric Hospital database. In order to develop a forecasting model, we used 4 years of data from January 2007 to December 2010 to construct the demand forecast model, whereas those from the following year (January to December 2011) were used to evaluate the model. Forecasted models were constructed with two ANN models: Radial Basis Function (RBF) and Multi-Layer Perceptron networks (MLP). The forecast accuracies for the models were evaluated via Mean Absolute Percentage Error (MAPE). The RBF was selected as the final model. The results demonstrated that monthly anxiety disorders patient visits can be predicted with good accuracy using the RBF model technique in time series analysis since the MAPE is below 20%. The majority of patients was female, married, farmers, aged between 40-59 years old and diagnosed with other anxiety disorders (F41). An average of one hundred and fifty patients of all ages attended each month at outpatient services with the highest being 244 and the lowest 76. The forecast cases exceeded the actual clinical cases in the 20-39 age groups. Accurate forecasting of outpatient visits can play a significant role in the management of a health care system.