遗传优化的广义回归神经网络在肺结核发病预测中的应用

目的 利用西安市2008年1月至2019年10月结核病月发病率数据分别建立广义回归神经网络和BP神经网络预测模型,提出利用遗传算法的全局搜索能力优化广义回归神经的光滑因子。 方法 以2008年1月至2018年12月发病率作为训练样本,以2019年1月至10月发病率作为测试样本,对两种模型的仿真预测结果进行对比分析。 结果 遗传优化的广义回归神经网络其预测的平均绝对误差（MAE）,均方根误差（RMSE）,平均相对误差（MAPE）均小于BP神经网络,预测效果更优。 结论 遗传优化的广义回归神经网络较BP神经网络在肺结核发病率预测中有更好的拟合效果和预测精度,其预测效果更理想。其具有良好的实用价值,为肺结核发病率的预测提供了一种有效的方法。

Application of genetic-optimized generalized regression neural network in the prediction of tuberculosis

To optimize the smoothing factor of the generalized regression neural network based on monthly incidence rate of tuberculosis from January 2008 to October 2019 in Xi’an, the prediction models of generalized regression neural network and BP neural network were established respectively. Methods The incidence rates from January 2008 to December 2018 were used as training samples, and the incidence rates from January to October 2019 as the test samples, comparing and analyzing the real prediction results of the two models. Results The results showed that the mean absolute error(MAE), root mean square error(RMSE) and mean relative error(MAPE) of the genetic-optimized generalized regression neural network were smaller than those of BP neural network, and the prediction effect was better. Conclusion The genetic-optimized generalized regression neural network has better fitting effect and prediction accuracy than BP neural network in the prediction of tuberculosis incidence rate, its prediction effect is more ideal. It has good practical value and provides an effective method for the prediction of the incidence of tuberculosis.