| Objectives:To standardize TCM symptom and syndrome terms clinical used through the guidance of TCM theory,and build a electronic medical record system of malignant tumor under the guidance of clinical experience using information technology.Then refer to the syndrome distribution research of small cell lung cancer.and combine symptoms and other related factors that may influence syndromes of patients with small cell lung cancer,establish intelligent syndrome differentiation model of small cell lung cancer using the machine learning,a algorithm of artificial intelligence.Methods:①On the basis of research on standardized TCM clinical symptom terms of malignant tumor,combined with the information of basic patient information,basic disease information,diagnosis and treatment process,related examination and underlying symptoms necessary for TCM clinical diagnosis,design the business framework of electronic medical record system for malignant tumor in line with the clinical habits of TCM oncologists.②Establish database structure using MySQL,then built the data dictionary of standard symptom and syndrome terms,and the relationship between those symptoms and syndromes,and users table,patient information table,disease information table,treatment table,examination table,and make the system having good logical and scalability in the aspect of data for further optimization research.To build the front-end framework of the system using Vue,which having a friendly operating interface that can adapt to different browsers and different screen sizes both on PC and mobile terminals.Node.js is used for data interaction to ensure system stability and data transmission efficiency.③Using the above electronic medical record system to collect diagnostic information of patients and export data.Then make system clustering analysis of those data using SPSS23.0 software to extract the TCM syndrome factor.We choose Phi4 point correlation method of system clustering because of those data belongs to binary classification variables.④Factors that may affect TCM syndromes of SCLC that have not been analyzed in existing studies include whether patients have undergone surgery,chemotherapy,radiotherapy,immunotherapy,targeted therapy,gender,age,smoking history,KPS score,tumor stage,distant metastasis and other factors related to syndromes.This kind of data belongs to four-cell table data,so chi-square test or Fisher’s exact probability method was used for statistical analysis.⑤The neural network model was constructed by using SPSS23.0 multilayer perceptron of neural network,using clinical symptoms,other influencing factors,and the clinical symptoms+other influencing factors as the characteristic factors,and the result of syndrome differentiation as the target variable.The accuracy of training set and test set and the area under receiver operating characteristic curve(ROC)were used to evaluate the model.Results:① The electronic medical record system for malignant tumor was built to collect patient information,assist TCM syndrome differentiation,review the course of disease and follow up in real time,and export data that can be directly analyzed by statistical software such as SPSS.② 286 cases of small cell lung cancer that met the criteria were collected by electronic medical record system of malignant tumor.A total of 182 symptoms appeared in the patients,and the top ten symptoms were yellow tongue coating,dark tongue,fatigue,purple tongue,angry pulse,cough,thin tongue coating,rapid pulse,thin pulse,loss of appetite.The top ten tongue and pulse were yellow tongue coating,dark tongue,purple tongue,angry pulse,thin tongue coating,rapid pulse,thin pulse,white tongue coating,red tongue and thick hypoglossal veins.Besides tongue and pulse,the top ten symptoms were fatigue,cough,loss of appetite,insomnia,shortness of breath,white sputum,chest tightness,dry mouth,dry stool,nausea.③ The symptoms of 286 patients with small cell lung cancer were grouped into 26 categories.Category 1:spontaneous sweating,night sweats,insomnia,pale lips and tongue,pale complexion,vomiting of phlegm-drool,back pain;Category 2:low fever,cold limbs,stabbing pain in the rib,back burning pain,lower limb pain;Category 3:hot flash,sticky discharge,shortness of breath,chest congestion,dull pain in the upper limb;Category 4:dysphoria in chestpalms-soles,the night fever,easy to catch a cold,lassitude of essence-spirit,fatigue,lethargy,dreaminess,delirious,forgetfulness,voice low,bradykinesia,language disorders,emotional depression,mental fatigue,irritability,sallow complexion,heavy body trapped,hearing loss,tinnitus,vision fatigue,decreased visual acuity,red eyes,sore gums or canker,bitter taste,dry mouth,thirst not to drink,fullness in the stomach,loss of appetite,hemiplegia,adverse limb movement,nausea,vomiting,nocturnal frequent urination,reduced stool times,weakness in defecation;Category 5:aversion to wind,migraine,itching in pharynx,phlegm difficult to cough up,loose stools,increased number of stools,back pain,upper limb pain;Category 6:fear of chills,runny nose,dry pharynx,dry mouth,nose,throat,lips and tongue,dry skin,red tongue edge,cold chest pain;Category 7:emasculation,dry eyes,hemoptysis,crooked mouth and eyes,pleural effusion,pericardial effusion,hematochezia,lower limb pain;Category 8:palpitations,bad breath,epigastric pain,lumbago pain;Category 9:unwillingness to speak,head colic,cough and asthma,edema,retching,short urine,weak pulse;Category 10:long sigh,dizziness,sore throat,mouth and tongue ulcer,tremor of limbs;Category 11:deafness,hoarseness,cracked tongue,tongue less fluid;Category 12:blurred vision,cough,cough sticky sputum,cough white sputum,cough sound heavy,urine yellow,bursting chest pain,bursting back pain,lumbago,upper limb pain;Category 13:sticky mouth,clear and thin sputum,soft waist and knees,unfavorable urination,painful urine,dry stools,chronic diarrhea,fat tongue,thick and greasy tongue coating,wandering pain in the ribs,stabbing pain in the lower limbs;Category 14:thirst and prefer cold drinks,abdominal distention,frequent urination,abdominal distention and pain,upper limb tingling;Category 15:tastelessness,teeth marks tongue,purple tongue,thin tongue coating,white tongue coating,angry pulse,rapid pulse,thin pulse,head pain;Category 16:thirst for water,less tongue coating,smooth pulse;Category 17:dry cough without sputum,head wandering pain,chest wandering pain,epigastric wandering pain,back wandering pain,abdominal wandering pain,upper limb wandering pain,lower limb wandering pain;Category 18:cough yellow phlegm,stomach distention,chest burning pain,epigastric distention pain;Category 19:throat phlegm,head dull pain,chest stabbing pain,epigastric stabbing pain;Category 20:hypochondriac discomfort,belching,limb cramps,head bursting pain,hypochondriac bursting pain;Category 21:stomach noise,belching,hiccup,limb numbness,sticky stool,more fart,pale tongue,irregular pulse,flank dull pain,waist dull pain;Category 22:carbuncle sore furuncle swelling,rash,itching of the skin,hand and foot peeling,dark tongue,red tongue,ecchymosis of the tongue,yellow tongue coating,thick hypoglossal veins,neck stabbing pain;Class 23:neck dull pain,back dull pain;Class 24:chest dull pain,abdominal dull pain,lower limb dull pain;Class 25:back cold pain;Class 26:waist stabbing pain.④Studies on the relationship between symptoms and factors related to syndromes such as whether patients have undergone surgery,chemotherapy,radiotherapy,immunotherapy and targeted therapy showed that:Gallbladder phlegm syndrome,kidney-Yang deficiency syndrome,gallbladder solid heat syndrome,Liver-yin deficiency syndrome and heart blood stasis syndrome showed difference between patients who have undergone surgery or not was statistically significant(P<0.05).Blood stasis syndrome,liver-yin deficiency syndrome,qi stagnation syndrome and blood heat syndrome showed difference between patients who have undergone chemotherapy or not was statistically significant(P<0.05).Lung solid heat syndrome and qi stagnation syndrome showed difference between patients who have undergone radiotherapy or not was statistically significant(P<0.05).Solid heat syndrome,deficiency of blood vessels syndrome,brain blood stasis and heart blood stasis syndrome showed difference between patients who have undergone immunotherapy or not was statistically significant(P<0.05).All the syndromes were not statistically significant between patients who have undergone targeted therapy or not(P>0.05).Studies on the relationship between other influencing factors and syndromes showed that syndromes of SCLC patients were statistically significant between different gender,age,smoking history,KPS score,tumor TNM stage,distant metastasis,etc.(P<0.05).⑤The neural network model was established with the clinical symptoms of small cell lung cancer as the characteristic factor and single syndrome combination as the target dependent variable.The model hidden layer contained 19 units.The accuracy of the training set was 95.7%and the accuracy of the test set was 93.4%.The hidden layer of the neural network model established with gender,age,smoking history,tumor TNM stage,distant metastasis,KPS score,surgery,chemotherapy,radiotherapy and immunotherapy as the characteristic factors,and single syndrome combination as the target dependent variable contained 17 units,and the accuracy of the training set was 91.4%and the accuracy of the test set was 90.3%.The hidden layer of the neural network model,which took the clinical symptoms of small-cell lung cancer combined with gender,age,smoking history,tumor TNM stage,distant metastasis,KPS score,surgery,chemotherapy,radiotherapy and immunotherapy as the characteristic factors,contained 18 units.The accuracy of the model training set was 95.0%,and the accuracy of the test set was 92.5%.Conclusions:①Electronic medical record system of malignant tumor can play a very good auxiliary role in clinical and scientific research,which can improve the efficiency of TCM clinical diagnosis and treatment,real-time follow-up,and provide high-quality scientific research data.②The symptoms of patients with small cell lung cancer are diverse,and most patients present showed systemic symptoms.Cluster analysis shows that the relationship between symptoms of SCLC patients is complex and there are many classification levels.Many of the 26 categories finally gathered can be divided into subcategories,and many of the categories contain different syndrome type information at the same time,while the distribution is not concentrated.Therefore,the traditional description of syndrome with four characters or eight characters cannot include all the information contained in these symptoms,and thus cannot reflect the whole picture of the patient’s condition.The extraction results of location syndrome elements and nature syndrome elements of disease showed a variety of syndrome factor is likely to appear in a same patient at the same time.Only through cluster analysis cannot get the combination of all the syndrome elements each patient presented.Therefore,in order to comprehensively describe the patient’s condition,it is necessary to completely describe the combination of patient syndrome and elements through the original symptom information recorded in the electronic medical record system combined with detailed diagnostic criteria and clinical experience formed by previous studies,so as to achieve more accurate diagnosis and treatment..③According to the study on the relationship between different treatment methods and syndromes,TCM syndrome of SCLC patients is related to surgery,chemotherapy,radiotherapy and immunological therapy,but not related to targeted therapy.According to the study on the relationship between other influencing factors and syndromes,TCM syndromes of SCLC patients is related to gender,age,smoking history,KPS score,tumor TNM stage,distant metastasis,but not related to VALG stage.④Artificial neural network algorithm can establish TCM intelligent syndrome differentiation model with high accuracy,providing reference for TCM clinical diagnosis of small cell lung cancer.The overall accuracy of the training set and test set of the three models established by taking TCM clinical symptoms,other influencing factors of syndrome type and TCM clinical symptoms+ other influencing factors of syndrome as characteristic factors were all greater than 90%.No matter whether the patient’s clinical symptoms are obvious or not,the neural network model can be used to get more accurate TCM syndrome differentiation. |
PDF Full Text Request