Research On Constructing Data Mining Model Of The TCM Five-viscera Correlation Of Myasthenia Gravis

ObjectiveMyasthenia gravis(MG) is a common disease of Neurology. The incidence is about 0.25～2.0/100 thousand, and about 15%～20% patients present articulo. If the patients have not been treated in time, they would die. The pathogenesis of MG is undefined, and most specialists presume that it is an autoimmune disease. MG is a worldwide difficult problem in medical science. The popular medicine and treatment in western medicine include corticoid, anticholinesterase, immunodepressant, immunoglobulin, plasmapheresis, actinotherapy, and exairesis. However, the side effect of western medical treatments is very serious, and the disease is easy to recrudescent and dependent. Professor Deng Tie Tao applys the "spleen deficiency and five-viscera related" theory in Traditional Chinese Medicine (TCM) treatment of MG, and the therapeutic effect is satisfactory.The TCM five-viscera correlation theory is proposed by famous professor Deng Tie Tao, which breaks the restrictions of the five elements theory. The TCM five-viscera correlation theory is about the physiologic function of five viscera systems, the pathological changing feature of five viscera systems, and the interrelation among systems. The five-viscera correlation theory is established based on TCM clinical practice. It is a clinical thinking pattern of TCM, and is a summary of complicated clinical phenomenon. The five-viscera correlation theory is known as the modern version of developing five elements theory. The five-viscera correlation theory is able to reveal the relationship of five viscera exactly, and accommodate the development of modern science and the innovation of TCM clinical practice. Correlation studies were focused on the qualitative exploration of theoretical origin, or data quantitative analysis based on the classical test theory. However, it was short of the studies that quantitative show of theoretical meaning, data mining of modern clinical information, and the quantitative validation between them.Based on MG, the Rough Set and Association Rules in data mining are applied during the study of symptoms and pathology, in order to construct data mining model of five-viscera correlation, find the association of five viscera systems, and create a mathematical method in studying five-viscera correlation.Methods1. Collected 567 cases of patients according to diagnosis, internalization, depletion standards of MG.2. Designed the information collection questionnaire of Chinese Medical four diagnostic methods, collected clinical information on patients with MG, filled out the questionnaire form.3. Established a information database of patients with MG, and preprocessed data4. Analyzed frequency statistics of general information of patients with MG using SPSS 17.0.5. Reduced the symptoms of every viscera system of MG patients, using the attribute reduction algorithm based on conditional entropy by MATLAB 2010B, to find the core symptoms. Mined the TCM five viscera correlation pathogenesis of MG using Association Rules algorithm of SQL SERVER 2005, to find out the mathematical relationship of five-viscera correlation and construct the data mining model of five-viscera correlation.Result1. Frequency statistical result(1) Incidence in patients with MG male/female=1/1.52, the incidence peak is young.(2) The diseases combined with MG are divided into the following categories: thymus diseases, endocrine diseases, muscle diseases, and other diseases.(3) The frequency distributions of clinical type were as follows:Ⅱ-B type 68.6%,Ⅱ-A type 13.8%,Ⅰtype 11.5%,Ⅳtype 3.5%,Ⅲtype 2.5%,Ⅴtype with muscular atrophy 0.2%.(4) The common symptoms of MG included:ptosis, eyelid fatigue, limb weakness, difficulty swallowing and chewing, diplopia, dysarthria, respiratory weakness, restricted eye movements, inability to raise neck etc symptoms of muscle weakness. the combined symptoms of five viscera, such as fatigue, poor appetite, bowel, soft knee weakness, sore waist, low back pain, insomnia, irritability, more saliva, blurred vision, sweating, loose stools, dizziness, chest distress, physical hardship weight, lazy words, stethocatharsis, palpitation, etc.(5) The descending frequency distributions of the muscles involved are that the eye muscle, the limb muscles, bulbar muscles, facial muscles, respiratory muscles, cervical muscle, and other muscles.(6) The statistics results of diseased viscera were as follows:spleen 567 cases, kidney 363 cases, liver 355 cases, lung 263 cases, heart 213 cases. The frequency distributions Involved multiple diseased viscera were as follows: the five viscera 88 cases, four viscera 136 cases, three viscera 143 cases, two viscera 129 cases, spleen involved alone 71 cases.2. The results of attribute reduction using Rough SetThe core symptom sets were as follows:kidney{difficulty swallowing and chewing, soft knee weakness, inability to raise neck, tinnitus, nocturia};heart{insomnia, sweating, palpitations, chest distress, easy to panic};lung{dysarthria, breathing weakness, cough, shortness of breath, low and hoarse voice};liver{restricted eye movements, diplopia, dizziness, irritability, blurred vision};The spleen was diseased constantly without negative value of decision attribute, and the symptoms set may not be reduced.3. The result of Association Rule(1) Found 3- frequence item sets.(2) Found the relationship between two viscera is bidirectional, which conformed to the five-viscera correlation theory.(3) Found the relationship between spleen and other viscera in descending order by importance were kidney, liver, lung, heart. And the bidirectional relationship between spleen and the other four viscera arose earliest, which conformed to the "spleen-oriented, five viscera related"(4) Found 50 relationship rules, including 30 three-viscera relationship rules and 20 two-viscera relationship rules. The two most important rules were the kidney, spleen→lung, kidney→lung.(5)Found the rules that spleen combining the other viscus forecasted the third viscus were more important than that two other viscera forecasted spleen, if three viscera were diseased simultaneously.Conelusion1. It is a coverted pattern from deductive reasoning to inductive reasoning, that the process of analyzing pathogenesis characteristics of TCM with mathematical method. It enhances deducibility and scientificalness.2. The core symptoms sets of five viscera solved by Entropy-based reduction algorithm met the clinical requirements of the actual syndrome differentiation.3. The association rules were used the first time in the correlation of five viscera, and found the spleen was the main involvement of MG, following with kidney, liver, lung, heart, with the progression of the disease, it appeared the correlation of spleen-kidney, spleen-liver, spleen-kidney-lung, spleen-kidney-heart, lung-spleen-heart, and so on. The result conformed to the basic pathogenesis characteristics "spleen-oriented, five viscera related".4. The five-viscera correlation theory were validated more clearly by use of Rough Sets and Association Rules, including syndrome differentiation and the relationship between two viscera. The results conformed to clinical practice, were simple and straightforward, and showed that the two methods for the mathematical study of five-viscera correlation theory was feasible.