中文摘要：

为制定人体红细胞计数参考值的统一标准提供科学依据，以健康老年女性为例，收集了30865例健康老年女性的红细胞计数参考值，用5层BP神经网络分析了地理环境的海拔高度、年目照时数、年目照百分率、年平均气温、气温年较差、年平均相对湿度、年降水量和年平均风速等要素对其的影响，通过不同的地理环境变量组合，以训练误差为标准，评估不同地理变量对其的影响。结果表明：海拔高度、气温、湿度和风速对红细胞计数参考值的影响比较显著。认为临床诊断中需要考虑长期在不同海拔高度、气温、湿度和风速环境条件下生活的人群红细胞计数参考值的差异性。

英文摘要：

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 discusses geographical influencing factors to normal reference value of red blood cell count of old women based on artificial neural network. Reference value of red blood cell count is connected to geographical environments by air condition, diet structure and settlement environment etc which interact with blood indirectly. To study the nonlinear relationship between them can help to build standard for reference value of red blood cell count based on geographical factors. In order to supply a basis for uniting the reference value of red blood cell count, a research is made about geographical influencing factors to normal reference value of red blood cell count. 8 geography factors are divided into 10 groups. After building 5 layers neural network to each group, training errors were obtained till neural network stopped. Important of geographical influence can be evaluation based on training errors. The result indicates that influence of altitude, temperature, humidity and air flow velocity to red blood cell count is obvious. If the geography factors known, the reference value of red blood cell count can be simulated by the neural network more accurately.