神经网络预测方法在抑郁症疗效研究中的应用探讨

Application of ANN prognosing method in curative effect research of depression

目的应用神经网络方法进行针刺治疗抑郁症疗效的深入研究。方法基于临床研究,应用神经网络方法进行疗效预测。关注与抑郁症针刺治疗效果密切相关的治疗因素和疾病状态指标,将其作为神经网络的输入变量,将经治疗后量表的评定结果作为输出变量。收集临床病例形成神经网络的训练和测试样本,并讨论其临床应用性。结果药物组和针刺组的临床总疗效之间无显著性差异;神经网络模型对HAMD、SDS量表评分的预测符合率分别为79.03%和73.93%。结论在符合针灸学科特点的临床研究中,神经网络方法的应用是一个值得研究的方向。

Objective To study the curative effect of acupuncture on depression through applying the prognosing method of artificial neural network （ANN）. Method The curative effect was prognosed by using ANN method based on clinical studies. The attentionwas paid to the therapeutical factors and pathological indexes, which closely related to the curative effect of acupuncture on depression. These factors and indexes were chosen as ANN input variables and the outcomes of scales after the treatment as the output variables. The collected cases were divided into the training samples and testing samples of ANN, and their clinical application were discussed. Result There was no significant difference in the total curative effects between two groups. The predictive rates of ANN methods to the scores from HAMD and SDS scales were respectively 79.03 % and 73.93 %. Conclusion The application of ANN method is one of valuable research objects in the clinical study of acupuncture science.