FORECASTING MEASLES COVERAGE USING ARTIFICIAL NEURAL NETWORK.

OBJECTIVES: To propose the forecasting model for monthly measles infant immunization coverage a significant concern in disease management and control. DESIGN: The reported data of monthly infant measles immunization coverage to National institute of health, Islamabad, Pakistan from January 2009 to October 2014 for the present study has been taken from Pakistan bureau of statistics with total time series entities 70. National institute of health, Islamabad took the record of per month number of doses administered (0-11 months) children by the registered health centre in Pakistan. PERIOD: January 2009 to October 2014. SETTING: Pakistan Bureau of Statistics (Statistics House) Methods: Artificial neural network (ANN) analysis has been carried out to develop a forecasting model. RESULTS: Several combinations of input and hidden layers are executed by taking under consideration the root mean square error in the selection of final efficient model. The efficient ANN model has twelve input nodes, nine hidden nodes and one output node with back propagation learning rate 0.05 and preferred activation function as hyperbolic tangent function. The established ANN model revealed that the increment for infant measles coverage is 7.58% expected in next six month. CONCLUSIONS: ANN 12-9-1 is an efficient model for forecasting the monthly measles infant immunization coverage in Pakistan. [ABSTRACT FROM AUTHOR]