Auxiliary Prediction Model Of AD Disease Condition Based On Improved MDEBM

Alzheimer ’ s disease is extremely harmful to middle-aged and elderly people,so the incidence rate has gradually increased in recent years.The disease is irreversible with a long course,and there is no effective treatment,so we can only adhere to the policy of early detection,early intervention,and early prognosis.Among these,its early screening is particularly important.The Multi-stage Event Discriminant Model（MDEBM）takes the changes of biomarker attributes as events,and obtains the multi-stage timeline of Alzheimer’s disease progression through sample cross-sectional data analysis,so as to realize the diagnosis of Alzheimer’s disease.Early screening of patients.However,the current model has the problems of training data dependence of event center sequence and limited input mode of the model.In this regard,the following work is carried out in this paper:1.In view of the deviation between the theoretical center sequence and the practical application center sequence generated by the cross-sectional data of the training dataset,this paper proposes a multi-stage event discrimination model（AMDEBM）based on attention mask.The corresponding center sequences generated for different batches of task samples are retained through the attention mask,and the importance saved in different masks is changed through training,and finally a new center sequence is obtained through the task-corresponding mask.This paper conducts experiments on the data of the ADNI dataset,and the accuracy of this model on the classification task is improved by 1.76%.2.A progressive AMDEBM disease staging prediction model based on MRI is proposed:neuroimaging data is added as input,which expands the input modal support of the model;for the negative induction problem generated by irrelevant data,the autoencoder is used as an expert gate pair.The input and task models are split,and finally the scalability and robustness of the improved model are verified through experiments on different public datasets.3.Designed and implemented the We Chat applet "Auxiliary diagnosis system for Alzheimer’s disease for middle-aged and elderly people",which includes diagnostic functions using different inputs from cognitive scales to neuroimaging data,from demand design,Functional design and performance analysis are described from multiple perspectives to meet the needs of middle-aged and elderly users.