伤寒副伤寒与气象及地质因素关系的BP神经网络模型研究

目的探讨桂林市伤寒副伤寒流行的气象地质因素，建立桂林市气象因素及地质因素与伤寒副伤寒发病的预测模型，将伤寒、副伤寒发病等级与影响因素进行训练预测，并评价模型的拟合效果。方法收集1994—2001年桂林辖区各县、市区报告的伤寒副伤寒疫情数据、同期桂林市平均降雨量、平均气温等气象资料、辖区各县36个乡镇的地质因素，利用Matlab6．5软件对人工神经网络BP模型进行构建、训练及模拟，并对模拟效果进行评价。结果气象因素和地质因素伤寒副伤寒发病预测平均误差率分别为1．21％和2．64％，决定系数砰分别为0．996和0．998。结论伤寒副伤寒与气象因素及地质因素关系的BP神经网络模拟合效果较好，有进一步研究的价值。

Model of Back- Propagation Neural Network About Meteorological/geological Factors and Typhoid Fever, Paratyphoid Fever

Objective To investigate the relations between meteorological/geological factors and typhoid fever and paratyphoid fever, Back-Propagation artificial neural network model was established and evaluated. Methods The data of incidence of typhoid fever and paratyphoid fever and meteorological / geological factors in Guilin city from 1994 to 2001 were collected. The model of back- Propagation artificial neural net work was built by Matlab version 6. 5. Results The MER of model meteorological and geological factors were 1.21% and 2. 64% , respectively. The R2 of model meteorological and geological factors were 0. 996 and 0. 998 respectively. Conclusions The BP Neural Network model has practical value to popularize in simulation of situation among meteorological/geological factors, typhoid fever and paratyphoid fever.