Causal Analysis Of Literature-based In Symptom Of Alzheimer's Disease

With the extensive application and popularization of the Internet,the big data era is coming,and a variety of analysis methods based on big data have been developed.This thesis tries to obtain some deep causative analysis conclusions that are related to pathology,physiology,behavior,sociology and other aspects about AD by literature data processing techniques,include the classification and feature extraction of Literature.This thesis conducts the classification and feature extraction of AD literatures,which consist of about 120,000 abstracts that are mainly published in various medical journals.This thesis establishes Bayesian causal map based on the data,and find out the main reason for the impact of AD through causality analysis,which could provide necessary data support for clinical trials.This thesis tries to use a literature-based disease analysis technique and applies statistical causal analysis to mining causal relationship from the literature data.At the same time,an improved naive Bayesian literature classification method is proposed and a compromise scheme is proposed which combines the hypothesis of word independence and the joint distribution of words in the literature classification.And then through the results of literature analysis,a new constraint of data constraints and causal calculations are established,which could reduce the size of data attributes for the analysis of the interference.Medical research places more emphasis on causality rather than association,which has been a problem between data analysts and clinicians.This thesis mines causal relationship from the pure data through causal analysis techniques,and based on the constructing of Bayesian network,a practical linear regression equation model is proposed,then the degree of direct causal impact between variables can be easily calculated.This shows the advantages and technical characteristics of statistical causality analysis purely from the literature data.