Bayes判别分析在基于儿茶酚胺的阿尔茨海默病预测研究中的应用

Diagnosis of Alzheimer's disease and its prediction basing on Bayes discriminant analysis method using Catecholamines data

利用判别分析方法通过对测量的12个阿尔茨海默氏症(Alzheimer's disease,AD)患者和12个健康者的尿液样本中儿茶酚胺(Catecholamines,CA)含量的测定及浓度作为训练集建立判别函数,进行疾病的诊断,交叉验证错误率为4.2%。采用随机余下的4个数据带入判别函数,进行预测,结果表明具有很好的预测能力,正确率达到了100%。此方法可以通过测量人体尿液中CA含量测定及浓度来诊断AD,对AD的尽早检测和早期治疗非常重要。2组线性判别函数分别为-19.91024+0.21873*E+0.23742*NE+0.11155*DOA+0.41789*L-DOPA-0.12661*DOPAC;-2.24864+0.03070*E+0.04914*NE+0 06892*DOA+0.01704*L-DOPA+0.01598*DOPAC。

In this paper, discriminate analysis method was applied to diagnose the Alzheimer＇s disease （AD） using the data of determination and concentration of the Catecholamines （CA） content in the urine samples. The measurement of 12 AD patients and 12 healthy individuals was applied to build the model as training set. The cross-validation error rate is 4.2%. The model has a good predictive ability with the correct rate of 100%. This method plays an important role for AD detection and early treatment through measuring CA content determination and concentration in human urine. The linear discriminate function was -19.91024＋0.21873＊E＋0.23742＊NE＋0.11155＊DOA＋0.41789＊L-DOPA-0.12661＊DOPAC;-2.24864＋0.03070＊E＋0.04914＊NE＋0 06892＊DOA＋0.01704＊L-DOPA＋0.01598＊DOPAC.