基于多检查结果融合的MCI进展预测方法

Converter prediction of MCI based on fusion of multiple examination results

为提高轻度认知障碍(MCI)患者向阿尔茨海默症(AD)阶段病情进展的预测性能,提出一种融合病人多项检查数据进行学习的半监督神经网络新型模型MVIDG。通过mRMR算法对高维特征进行降维,对病人单项检查数据使用Dual-GCN进行基础模型训练,利用改进后的MVCDN网络对各项检查数据训练出的模型进行融合,以对未来一年内病人从MCI阶段向AD阶段的病情进展进行预测。实验结果表明,所提模型可有效整合病人多项检查结果以提高预测性能,效果优于其它数据融合方法。

To improve the prediction performance of patients with mild cognitive impairment(MCI)to the stage of Alzheimer’s disease(AD),a semi-supervised neural network model MVIDG was proposed that integrated multiple examination data of patients for learning.The dimensionality reduction of high-dimensional features was carried out through the mRMR algorithm,and the basic model training was performed on the patient’s single inspection data using Dual-GCN,and the improved MVCDN network was used to fuse the models trained by each inspection data,to predict the patient’s disease progression from MCI stage to AD stage in the next year.Experimental results show that the proposed model can effectively integrate multiple patient examination results to improve prediction performance,and the effect is better than that of other data fusion methods.