神经网络方法在抑郁症早期筛查中应用的可行性

Application feasibility of ANN method in depression early screening

目的:探讨神经网络方法在抑郁症早期筛查中应用的可行性。方法:应用神经网络方法构建抑郁症早期筛查模型,收集高校学生和企业员工两个人群的抑郁症早期筛查数据形成神经网络训练和测试样本,筛选与抑郁症发病密切相关的背景因素,将其作为神经网络的输入变量,将抑郁自评量表评分作为输出变量,分析其临床应用的可行性。结果:神经网络模型对两个人群的量表评分的预测符合率分别为76.03%和73.93%。结论:将神经网络方法应用于抑郁症早期筛查具有较大潜力,值得深入研究。

Objective： To investigate application feasibility of ANN method in depression early screening. Methods： The depression early screening models were established by using the artificial neural network（ ANN） method. The survey data of college students and enterprise employees were collected as the ANN training and test specimens. The pathogenic factors closely related to the incidence of depression were chosen as ANN input variables and the output variables were the scores of the self-rating depression scale.Then,the application feasibility of ANN method was discussed. Results： The predictive coincidence rates of ANN models to the scores of HAMD and SDS scales of the two groups were 79.03 % and 73.93 %,respectively. Conclusions： The ANN method has a great potential for depression early screening,which needs further studies.