为了构建辅助诊断模型，以提高抑郁症诊断的准确率。在连续的阈值空间（8％～32％）内构建所有被试的功能脑网络并使用复杂网络理论对抑郁症患者的脑网络进行分析。通过设定阈值，根据统计显著性提取不同数量的节点属性与全局属性组合作为分类特征，并选择四种不同的分类算法进行分类研究，以得到构建一个准确率较高的模型。结果是SVM和神经网络算法在阈值P为0．05下，所建的模型的分类模型的准确率较高，分别达82．78％及81．36％，因此利用该方法所构建的诊断模型可以用于抑郁症的辅助临床诊断中。%In order to construct a computer-aided diagnosis model to improve the accuracy of depression diagnosis, we construct within continuous threshold space ( 8% ~32%) the functional brain networks which are all in testing and analyse the brain networks of the depressive patients with complicated network theory.By setting the threshold, we extract the node attributes in different numbers depending on the statistical significance and combine them with global attributes to be the classification features, and select four classification algorithms to carry out the classification research so as to build a model with higher accuracy.The result is that the classification model of the models built by SVM and neural network algorithm under the threshold P =0.05 has higher accuracy, reaches 82.78%and 81.36%respectively, so the diagnosis model built by this method can be used in computer-aided clinical diagnosis for depression.
展开▼
