中文摘要：

高敏C反应蛋白（hs-CRP）参考值制定时忽略了地理因素的影响，为建立更全面的制定标准，分析了中国健康成年人hs-CRP参考值与地理因素的关系。首先收集中国84个市（县）8 350例健康成年人的hs-CRP参考值，分析其与对应的23项地理指标的相关性，提取出其中存在显著相关性的5项地理指标。然后用hs-CRP参考值与提取的5项地理指标建立支持向量回归机模型，并用该模型预测出全国2 322个观测点健康成年人的hs-CRP参考值。最后运用ArcGIS软件进行地统计分析，经析取克里格插值，绘制中国健康成年人高敏C反应蛋白参考值地理分布图。研究表明：中国健康成年人的hs-CRP参考值与海拔高度、年均相对湿度、年均降水量、气温年较差、年均风速5项地理指标存在显著相关关系；支持向量回归机模型在此案例中有优秀的拟合预测能力；克里金插值法将点变成面，绘制出直观的hs-CRP参考值的地理分布趋势图；中国整体hs-CRP参考值呈现明显的从西北向东南递减的趋势。

英文摘要：

High sensitive C-reactive protein （hs-CRP） is an important predictor of cardiovascular and cerebrovascular diseases. In the dynamic balance human-nature system, the residents’ physical functions differ from each other in different regions. However, the geographical factors are neglected while establishing the reference value of hs-CRP. It will lead to inaccurate clinical diagnosis while using the same standard system of reference values of hs-CRP on the residents of different areas. Therefore, in this paper the relationship between reference value of hs-CRP and geographical factors was analyzed in order to establish a more comprehensive standard of reference values of hs-CRP. Firstly, by collecting the observed hs-CRP values of 8 350 Chinese healthy adults from 84 cities in China, the correlation analysis method was adopted to investigate the relationship between the reference value and 23 geographical factors with SPSS 21.0. Then the five geographical factors that were significantly correlated with the reference values of hs-CRP were extracted to perform the Support Vector Machine Regression model （SVR） with the reference values of hs-CRP. Good fitting of the model was obtained in this case, and reference values of hs-CRP of 2 322 cities in China were predicted by using the model. Finally, ArcGIS 10.0 was used to make Kriging interpolation with the predicted data and form the SVR model to produce the geographic distribution map of the reference values of hs-CRP of healthy Chinese. The results show that the geographical environment has an important effect on the hs-CRP reference value, and the reference value of hs-CRP is significantly correlated with 5 indexes, namely, the altitude, the average relative humidity, the annual average precipitation, the annual temperature range and the annual average wind speed. The Support Vector Machine Regression model got good fitting effect in this case with a small prediction error. And the extraction Kriging interpolation model also obtained good prediction accurac