Syndrome Element Differentiation Methodology Based On Data Mining Technology

As the characteristic and essence of the traditional Chinese medicine, the treatment after syndrome differentiation is a binding principle in the medical practices.Its scientific nature,excellence,and necessity have been proved by extensive medical practices in the history.Syndrome differentiation provides a particular treatment according to the individual patient's particular symptoms in a flexible mode,no matter whether the type of disorder has been confirmed or not.However,in the past ages,insufficient effort has been devoted to the study of the rules and systems of differentiation in the TCM circle.As there were various differentiation methods,confusing concepts,thousands of syndrome descriptions and implicit syndrome names,medical practitioners could not merely adopt a single method for syndrome differentiation,but have to apply a number of methods comprehensively.That brought large difficulties in medical practices,research and education.The establishment of uniform syndrome differentiation architecture doubtlessly laid a foundation for TCM's modernization.As a highly scientific methodology,the essential of the uniform syndrome differentiation architecture is the element differentiation.In the element differentiation,the process was composed of two steps-element reasoning and the forming of the syndrome name.The syndrome element reasoning is to be conducted according to the patient's symptoms and signs to pin down relevant syndrome elements,then combine elements to form a syndrome name that is in accordance to the TCM's accepted practices,so as to guide the practical prescription and treatment.Establishing scientific qualitative and quantitative criteria for the element differentiation under the uniform syndrome differentiation architecture is demanded not only by the TCM practitioners,but also as a foundation of the new medical care system demanded by the information society.In current practices of element differentiation,TCM practitioners normally acquire knowledge about syndrome elements recognition through collecting and distilling information from relevant publications and comments of experts.As most of the records in the publications on syndrome differentiation cover mostly subjective and qualitative methods,the treatment practices based on syndrome differentiation was comparatively empirical and subjective.Thus the syndrome differentiation knowledge acquired from publications and experts' comments were normally ambiguous and inconsistent.Therefore,setting up a scientific qualitative and quantitative measurement standard for syndrome element differentiation is the corner stone for promoting the application of the uniform syndrome differentiation architecture.Since the emergence of the uniform syndrome differentiation,researchers,facilitated by use medical databases,have made a lot of explorative work on fields such as acquisition of syndrome element differentiation rules,quantitative analysis on the impact of symptoms to syndrome elements and so on.Nevertheless,due to the defects and constraints of the learning methods,such as low deduction and high standard for empirical knowledge acquisition,there has been no breakthrough development in the research.Taking the information acquired in medical practices as research objective and employing data mining technologies,the research aims to explore for scientific methods of syndrome element recognition so as to provide technical support for finding a high speed and highly effective syndrome differentiation model for diseases prevention and treatment, and for solving some key technical issues in medical practices.Furthermore,the research aims to lay a foundation for convergence between the TCM and some inter-disciplinary studies such as the electronic medical record and the decision support system for diagnosis etc;and willlay a foundation for the studies on optimal disease treatment rules,optimal TCM prescription and specific therapy or medicine for a variety of complications.The research involves in using well acknowledged new data mining methods-some algorithms that support the SVM and the rough sets to conduct experimental analysis to collected data from medical cases,and hence to discuss on the feasibility of using those methods to establish a model for syndrome element differentiation in TCM practices.Meanwhile,the research will conduct experimental analysis on the double layer spectrum weighted cross cutting algorithm for syndrome element differentiation put forwarded by Prof.Zhu Wenfeng after he studied and distilled for scores of years.At last,a comparative analysis will be conducted to the results of the experiments.The problems to address are exploring for methods to improve the accuracy of the syndrome element differentiation and construct a platform for its fundamental research.The thesis begins with introductive summary to the research on objectivity of syndrome element differentiation.It follows with a summary of existing research techniques and a comparative analysis to them,and points out their defects,and hence put forward the advantages of applying fought set and SVM techniques.The paper then discusses the relationship between the data mining and studies on TCM diagnosis,acquisition of the rules for association between rough set and syndrome element differentiation,SVM and construction of a predictive model for syndrome element differentiation,the double layer spectrum weighted cross cutting algorithm,and construction of syndrome element table.In conclusion,a summary was made to wrap up the discussion and a direction was pointed out for future studies.