| As a very common and incurable senile disease,Alzheimer’s disease seriously affects the physical and mental health,work and life of the elderly.Because the early stage of Alzheimer’s disease is not obvious,it is often mistaken for normal aging of the human body and the best treatment stage is missed.Early detection and treatment intervention is the key to delay the progression of the disease.In this thesis,a model of data preprocessing and computer aided diagnosis for Alzheimer’s disease was proposed based on expert knowledge.There are dirty data in clinical data for many reasons,so data preprocessing before auxiliary diagnostic modeling is indispensable.As for the missing data in the data set,this thesis designs and implements the missing data interpolation method of mixed interpolation according to columns by combining the four methods of mean interpolation,regression interpolation,support vecto r machine interpolation and multiple interpolation.By comparing the effects of the mixed interpolation method with the above four interpolation methods and giving the comparison results,the experiment shows that the results of the mixed interpolation method under different data missing rates have better performance in terms of root mean square error,mean absolute error and error rate,which proves the effectiveness of the interpolation mechanism.In addition,in the processing of missing data,a combination of deletion method and interpolation method is adopted and expert knowledge is referred.Compared with the direct interpolation method,the data set obtained by this method is more accurate.On the basis of missing data processing,combined with data transformation,outlier detection,sample unbalance processing,feature selection and expert knowledge constraints,a complete data pretreatment process is formed,which provides more accurate data support for subsequent auxiliary diagnostic modeling.In the clinical auxiliary diagnosis of Alzheimer’s disease,the scientific nature of the model should be considered in addition to the accuracy.For the diversity of features in Alzheimer’s disease data,this thesis proposes a Bayesian network modeling method of feature classification modeling and combination.The characteristics of the data set were classified according to medical knowledge,and each type of characteristics were modeled by Bayesian network,and then the complete Bayesian network model was obtained by weight allocation.The experimental results show that the Bayesian network feature classification modeling and combination method can improve the accuracy of the model,and most of the relevant features are taken into account in the model.In addit ion,the visualization and modifiability of Bayesian network make the model more scientific. |
