Study On Optimization Of Syndrome Differentiation Algorithm For Primary Insomnia

Research Objective:Taking the syndrome differentiation process of primary insomnia as the research object,according to the theory of "symptoms refer to abnormal life phenomena reflecting its connotation by etiology,location,nature and condition,i.e.’position potentia["[12],the 46 differentiation factors and weights established earlier by our team were optimized,so as to improve the accuracy of syndrome differentiation of primary insomnia in the whole system.This research is part of the "Study on the Development of TCM Artificial Intelligence Diagnosis and Treatment System "(NO:2018SZ0065)from March 20 to September 1919,a major project of the Science and Technology Department of Sichuan Province.It was also created in 2003 by Professor Yang Dianxing and Professor Peng Mingde.A continuation study based on the study of intelligent TCM syndrome differentiation and treatment system based on multidimensional spatial mathematical model.Research Method:(1)To optimize the 46 factors of syndrome differentiation for primary insomnia.By consulting the literature of primary insomnia in the past 20 years of China HowNet,358 articles were included,195 cases of primary insomnia with complete medical records were collected,and 103 cases of primary insomnia were collected by filling out forms in the outpatient department of famous Medical Museum of Chengdu University of Traditional Chinese Medicine from November 2017 to November 2018.Frequency statistics,factor analysis and principal component scores were used.Methods The symptoms and syndromes in literature and clinical data were analyzed respectively.According to the results,46 factors of syndrome differentiation were optimized for all diseases,and 21 factors of syndrome differentiation for primary insomnia were obtained.(2)Optimizing the weights of syndrome differentiation factors.According to the analysis results of load coefficient and frequency of symptoms in the first part of common factors,21-bit syndrome differentiation factors were coded.Through training samples and machine learning based on Y(ki)=(?)Si÷(?),the initial weights of syndrome differentiation factors and syndrome types were obtained.Then,according to the understanding of insomnia in TCM,the deviation of initial weights of syndrome types was judged,and the weights of symptoms and syndrome differentiation factors were adjusted seven times and machine science was used.The final weights of 21 syndrome differentiation factors-syndromes and symptoms-21 syndrome differentiation factors were obtained.(3)Testing the accuracy of the optimized algorithm model.From December 2018 to March 2019,32 cases of primary insomnia patients collected from Acupuncture and Moxibustion College of Affiliated Hospital of Chengdu University of Traditional Chinese Medicine to Zhentang Teaching Clinic and 762 cases collected from Sichuan Large Data Platform for TCM Syndrome Differentiation and Treatment were used as test samples.Symptoms in the samples were input into the software one by one(the software was based on the optimized algorithm model).Evidence types and clinical results of intelligent selection of the software were compared.If it is consistent,it will be accurate,otherwise it will be inaccurate.Research Result:(1)Optimizing 46 syndrome differentiation factors,21 syndrome differentiation factors(PI-d-21)suitable for primary insomnia syndrome differentiation were obtained:heart,god,liver,gallbladder,spleen,stomach,kidney,internal wind,impotence,blood stasis,qi stagnation,qi deficiency,blood deficiency,yin deficiency,yang Virtual,food,cold,hot,sputum,wet,drink.(2)Regulating the weight of symptoms-syndrome differentiation factors,we can get the final weight of syndrome differentiation factors-syndrome types,among which the weight of syndrome differentiation factors heart and spirit is high;the weight of syndrome differentiation factors stomach in phlegm-heat internal disturbance syndrome is greater than that in liver-depression stomach syndrome;the weight of syndrome differentiation factors drink and blood stasis in syndrome types is low;the weight of other syndrome differentiation factors in syndrome types reflects the pathological degree of syndrome types in primary insomnia.(3)According to the initial weights of symptoms-21 syndrome differentiation factors,the final weights of 21 syndrome differentiation factors-syndromes were obtained after machine learning and 7 times of optimization.Finally,the symptoms-syndromes differentiation algorithm model of primary insomnia was constructed.(4)The optimized symptoms-syndromes syndrome differentiation algorithm model has an accuracy rate of 92.19%for primary insomnia.Research Conclusion:(1)Twenty-one syndrome differentiation factors based on position potential can explain most of the syndromes of primary insomnia.(2)The optimized symptoms-syndromes syndrome differentiation algorithm model improves the accuracy of syndrome differentiation by 8.57%for primary insomnia.