Application Research Of TCM Syndrome Classification Of Gastroesophageal Reflux Disease Based On Machine Learning

ObjectiveGastroesophageal reflux disease(GERD)is a common and frequently-occurring disease of the digestive system.With the in-depth study of epidemiological characteristics,typical symptoms and quality of life,the exploration of pathogenesis and treatment methods has become a hot topic.In view of the characteristics of easy recurrence,refractory and long period of the disease,using traditional Chinese medicine theory to guide the use of drugs,the clinical efficacy is significant.However,the essence of Traditional Chinese medicine bases on treating accords to syndrome differentiation.Which comes down to the discrimination of TCM syndrome.The accuracy of syndrome differentiation is the guarantee of clinical efficacy.However,the symptoms are difficult to find and overlap,which increases difficulty of syndrome differentiation and subjects to subjective constraints,leading to differences in the results of syndrome differentiation.Therefore,this study takes GERD as the entry point,drawing on the concept of evidence-based medicine and using statistical methods to study the TCM syndromes,syndrome distribution and epidemiological characteristics of GERD.Then we can get the characteristics of the syndrome distribution of the disease by the methods above.On this basis,the clinical cases of liver-stomach-stagnancy-heat pattern and spleen and stomach deficiency and counter-flowing qi syndrome were screened.The GERD intelligent syndrome differentiation model was constructed by using Support Vector Machine,Neural Networks and Autoencoder to compare the accuracy of syndrome prediction.MethodsThis thesis includes literature review and clinical research.(1)The literature review is divided into three parts.The first part discusses the understanding of gastro-esophageal reflux disease in traditional Chinese medicine,which mainly discussing the names,pathogenesis,syndrome differentiation and treatment of Traditional Chinese medicine.The second part discusses the progress in diagnosis and treatment of Gastro-esophageal reflux disease in modern medicine,mainly includes the epidemiology,etiology,pathogenesis,diagnosis and treatment of gastroesophageal reflux disease.The third part discusses the application of machine learning technology in the research of TCM syndrome,which focusing on the machine Learning from the application of TCM syndrome research and the progress of TCM intelligence research.(2)The second part is clinical trial.Firstly,collecting 233 cases from Dongzhimen Hospital of Beijing University of Chinese Medicine based on the unified TCM syndrome scale.Second,importing the general information and the symptoms collected into excel.Then using the cluster analysis and factor to discuss the common rules of TCM syndrome,disease and syndrome distribution in patients with gastroesophageal reflux disease.Based on SPSS 25.0 statistical software.Finally,selecting a total of 98 samples from the liver-stomach-stagnancy-heat pattern and spleen and stomach deficiency and counter-flowing qi syndrome.Among them,selecting 49 samples in each of the two categories.,By according to the 1:1 ratio between categories,70% of all data were used as training sets and 30% were used as tests.The GERD intelligent dialectical model was constructed by Support Vector Machine(SVM),Neural Networks(NNs)and Autoencoder to compare the accuracy of syndrome prediction.Results(1)A total of 233 patients with GERD were enrolled in the study,including 108 males(46.4%)and 125 females(53.6%).The male to female ratio was 0.86:1.Among all patients,the minimum age was 20 years old and the maximum age was 79 years old.The mean age was 52.35±12.33 years old,age was related to gender distribution(P <0.05).167 cases(71.7%)were reflux esophagitis,62 cases(26.6%)were non-erosive reflux disease,and 4 cases were Barrett's esophagus(1.7%).State agencies have more staff(29.5%),winter(28.2%),and spring(33.8%).Diet,emotions,and climate are the causes of the disease.The diet is mostly sweet(16.7%),spicy(12.9%),and fast eating(10.7%).Emotional were irritability(38.2%),anxiety(29.2%)Worry(29.2%).The number of people with GerdQ score ?8(72.1%)is much higher than the number of people with <8 points(27.9%).The sleep status of the patients was still acceptable(91.4%),which could not find the correlation with the disease and its subtypes.(2)In the correlation study,Hp infection and BMI index were correlated with disease subtype distribution(P <0.05).(3)The syndromes of gastroesophageal reflux disease mainly include the following ten types: liver-stomach-stagnancy-heat pattern,syndrome of stagnated heat of gallbladder channel,syndrome of disharmony of gallbladder and stomach,syndrome of excessive heat in stomach fire,syndrome of incoordination between the liver and spleen,syndrome of dampnessheat of spleen and stomach,phlegm blockage with qi stagnation syndrome,spleen dysfunction syndrome,spleen and stomach deficiency and counter-flowing qi syndrome,syndrome of qi deficiency of lung and spleen.The location factors were spleen,stomach,liver,gallbladder and lungs.The disease syndrome is confirmed by qi stagnation and damp heat and heat.The deficiency syndrome is mainly qi deficiency,yang deficiency and yin deficiency.The key of the pathogenesis is deficiency,stagnation,dampness and heat.(4)In 98 cases of GERD,there were significant differences in cold,fatigue,backache,phlegm-fluid retention in stomach,ungratifying defecation,chest oppression,irritability,wiry pulse among liver-stomach-stagnancy-heat pattern and spleen and stomach deficiency and counter-flowing qi syndrome.(P < 0.05).(5)With the same training and test sample data,the accuracy rates of SVM,NNs and Autoencoder for the identification of gastroesophageal reflux disease,liver and stomach stagnation syndrome and spleen and stomach deficiency and counter-flowing Qi syndrome were 78.3% and 79.2% and 79.2% respectively.Conclusion(1)By using cluster analysis and factor analysis,the TCM syndromes of GERD can be summarized into the following ten types,which refers to liver-stomach-stagnancy-heat pattern,syndrome of stagnated heat of gallbladder channel,syndrome of disharmony of gallbladder and stomach,syndrome of excessive heat in stomach fire,syndrome of incoordination between the liver and spleen,syndrome of dampness-heat of spleen and stomach,phlegm blockage with qi stagnation syndrome,spleen dysfunction syndrome,spleen and stomach deficiency and counter-flowing qi syndrome,syndrome of qi deficiency of lung and spleen.The above methods objectively and truly reflect the distribution of syndromes and the characteristics of syndromes elements of GERD,and provide a reference for the diagnosis and treatment of GERD.(2)NNs and NNS+Autoencoder model are effective on diagnosis and predictive ability.On the methodology it is feasible to use Machine Learning establish TCM Intelligent Syndrome Differentiation Model of GERD.