The Application of Artificial Neural Network in the forecasting on Incidence of a Disease

The main objective of this paper is to discuss the meteorological factors affecting the incidence of hypertension and set up the forecasting model. Firstly, on the basis of statistical analysis, selection of main meteorological factors remarkably affecting hypertension is conducted for Yinchuan area. The factors, including average humidity, temperature swing of 48hous, daily temperature range and air pressure, as input variables, are used for studying and training of multilevel feed-forward neural network BP algorithm and an ANN hypertension model is developed for forecasting this disease. Results are follows: The ANN model structure is 4-14-1, that is, 4 input notes, 14 hidden notes and 1 output note. The training precision is 0.005 and the final error is 0.0048992 after 46 training steps. The simulative rate of ANN model and statistical model of same level are 62.4% and 47.7%, respectively. The forecasting rate of ANN model and statistical model of same level are 58.2% and 50.5%, respectively. The MAPE, MSE and MAE of ANN model are 25.2%, 21.0% and 16.2%, respectively, which are much smaller than statistical model. The method is easy to be finished by smaller error and higher ability on historical simulation and independent prediction, which provides a new method for forecasting the incidence of a disease.