Research On Association Rule Of Symptom, Syndrome And Drug Of Chinese Medicine In Chronic Glomerulonephritis

Chronic Glomerulonephritis is a great threat to human health as one major disease of the Chronic Kidney Disease(CKD). Meanwhile, the Traditional Chinese Medicine(TCM) is a medical science that adjust the function of the human body and it plays an irreplaceable role in the treatment of Chronic Glomerulonephritis. Determining the Symptom-Syndrome-Chinese Medicine(Herb) relationship is an important step in the process of TCM treatment. Through collecting the information about the patient’s symptoms and making effective assistant results based on Symptom-Syndrome-Herb association rules, it can help doctor to make medical dicison that is the main content of this thesis.The thesis will apply non-subset inspection policy in the process of mining association rule. And it adds an evaluation function based on non-core herb problem. Then it put forward an improved Symptom- Syndrome- Herb rule generating program that will be a rule base for the next step. At last, it accomplishes a TCM decision support functions. By comparing the experimental data, the proposed scheme shows its effectiveness and availability. Specific work are as follows:1. Imported improved non-subset inspection closed frequent itemsets mining and non-redundant rule generation algorithm. To reduce the size of item sets, used the Cherry Item search strategy in the preceding paragraph generating closed frequent item sets. To reduce redundancy of rules, combined with the Symptom-Syndrome-Herb characteristics and defined the redundancy rules to screen the rules. Testing on the CNKI Chinese literature test data sets, the algorithm compared to non-closed frequent item sets on the item sets size has an average compression ratio of 30.45%.2. For that some Chinese Medicine there are more frequent, but did not play a major role. Imported evaluation function of Herb and Syndrome correlation to retain more relevant Herb and Syndrome rule by setting the the threshold value. The frequency of occurrence decreased from 14.42% to 4.63% as the test result.3. Achieved the Chinese medical decision support functions. Because of the different representations of the same symptoms, setted the Chinese medical symptoms hierarchical coding unity. To improve the efficiency of input, used the collaborative filtering bases on symptom-based clinical treatment database to help doctor entry symptom. Built rule base through mining Symptom-Syndrome-Herb rules, and entered symptoms to vote the most weighted and matched rule. Then produce the result to assist doctors make decision. The result of coverage is 18.87%, absolute accuracy is 30%.