随着信息时代的飞跃发展，数据的维度越来越高，非结构化的数据量众多，特征提取和模式分类以其独特的优势吸引了众多研究者的兴趣，一些学者将其应用在了阿尔茨海默症的诊断和识别中。以往对于阿尔茨海默症进行识别的研究大多是基于核磁共振的脑成像数据进行的，而本研究使用的是的fNIRS(功能性近红外光谱成像)技术，相比于核磁共振，它具有安全、价格低廉、应用性更强的优点。本文借助该技术收集到了25个患有阿尔茨海默症的老年人和34个正常老年人处于静息状态时大脑的含氧血红蛋白数据和脱氧血红蛋白数据，利用MATLAB软件对每个人分别基于这两类数据构建了大脑不同区域之间功能活动的皮尔逊相关系数网络，之后设定阈值将其转变为二值网络。之后每个网络经计算得到了35个拓扑结构属性包括集聚系数、特征路径长度等，将这些属性构造成维度为35的特征空间。最后使用R软件中的e1071、randomForest工具包利用支持向量机算法和随机森林算法构造了分类器，利用留一交叉检验的方法计算了分类的准确度、特异度、灵敏度。对于这两种算法来说，分类结果显示基于脱氧血红蛋白进行分类的效果总是优于含氧血红蛋白的，最后得出了脱氧血红蛋白可以作为识别阿尔茨海默症的参考指标这一结论。

With the development of science and technology , the dimension of data is higher and higher, unstructured data is numerous.In this information era, feature extraction and pattern classification has attracted many researchers with its unique advantage and gradually applied in diagnosing patients. However, most of the existing research focuses on the use of magnetic resonance imaging data. In this study, a new technology which called fNIRS that is cheaper and safer than magnetic resonance imaging technology was used.The oxyhemoglobin and deoxyhemoglobin of 25 elderly people with alzheimer's disease and 34 normal elderly people were studied with this technology. MATLAB was used to build the Pearson correlation coefficient network based on the two types of data respectively, the e1071 and the randomForest toolboxs in R were used to make a classifier. The classification accuracy、sensitivity and specificity were evaluated using leave-one-out cross-validation. Applying the approach to the two types of datase led to a classification result that the deoxyhemoglobin is always better than oxyhemoglobin, so we made a conclusion that the deoxyhemoglobin can be used as an important reference for the recognition of alzheimer's disease.