基于神经网络的老年女性红细胞计数参考值与地理环境

Relationship between normal reference value of red blood cell count of old women and geographical factors based on NN

提出用神经网络来研究地理环境与红细胞计数参考值的关系。为制定红细胞计数参考值的统一标准提供科学依据,以健康老年女性红细胞计数参考值为例,运用BP神经网络分析了其与地理环境的海拔高度(X1)、年日照时数(X2)、年日照百分率(X3)、年平均气温(X4)、气温年较差(X5)、年平均相对湿度(X6)、年降水量(X7)、年平均风速(X8)等要素间复杂的非线性关系特征,通过5层神经网络,经过100次自学习建立模拟模型后,用此模型很好地模拟了健康老年女性红细胞计数参考值与地理环境的关系。如果知道了中国某地的地理要素,就可以用此模型估算这个地区的老年女性红细胞计数参考值。基于神经网络预测红细胞计数参考值的模型,模型自动定义具体的结构和参数,减少了预测红细胞计数参考值的人为性,计算方便准确。

Accurate, reliable reference values are essential for effective clinical evaluation and monitoring. Reference value of red blood cell count is a very important index for clinical evaluation. In order to provide united standard for reference value of red blood cell count, this paper discuss the nonlinear relationship between reference values of red blood cell count of old women and geography environments based on artificial neural network. Reference value of red blood cell count is connected with geography environments by atmospheric condition, diet structure and settlement environment etc which interact with blood indirectly. Studying the nonlinear relationship between them can help to build standard for reference value of red blood cell count based on geography factors. It is artificial neural network that have the ability to obtain the nonlinear relationship between variables. In order to supply a basis for uniting the reference value of red blood cell count, a research is made about the nonlinear relationship between the reference value of red blood cell count and eight geographical factors based on artificial neural network. After building 5 layers NN, the relationship was simulated by artificial neural network. After training 100 times, the network works better. If the geography factors were known, the reference value of red blood cell count can be simulated by the neural network more accurately.