| 1 ObjectiveThe purpose of this study is to explore Professor Piao Bingkui’s unique treatment based on pattern differentiation of lung cancer through machine learning,guided by the clinical thinking model of "disease-symptom-evidence-formulation-medication",in order to provide a reliable basis for the comprehensive transmission of Prof.Piao Bingkui’s characteristic experience and academic views in the treatment of lung cancer.2 Method2.1 Data collectionRelying on the case collection template formed by the National Natural Science Foundation of China Project "The Optimization Study of Cognitive Model based on Prototype Category Theory from Famous Veteran Doctors of TCM in Lung Cancer Diagnosis and Treatment"(No:81873399),we developed a medical case information collection catalog containing entry numbers and entry contents.Then Microsoft Office Excel 16.9.1(23011600)software was applied to standardize the collection of outpatient cases of lung cancer treated by Prof.Piao Bingkui on an entry-by-entry basis,and a total of 1325 cases were collected and organized.2.2 Data cleaningThe descriptions of symptoms,tongue coating,pulse conditions,western diagnosis,past history,TCM diagnosis,TCM evidence,drugs,evidence,and treatment were standardized according to "Differential Diagnosis of Chinese Medicine Symptoms","Diagnosis of Chinese Medicine",the International Classification of Diseases,Eleventh Revision(ICD-11),"Classification and Coding of Chinese Medicine Evidence"(GB/T15657-2021),"Pharmacopoeia of the People’s Republic of China",the 2020 edition of "Evidence Identification" and "Chinese Medicine Clinical Diagnosis Terminology Part 3:Treatment"(GB/T 16751.3-2021)2.3 Data analysis2.3.1 Descriptive statistical methodsThis study used Microsoft Office Excel 16.9.1(23011600)software to perform descriptive statistical analysis of basic patient information such as patient gender and age and information on Western medicine diagnosis,Chinese medicine diagnosis,pattern diagnosis,and prescribed medications in the medical case data set.2.3.2 Construction of a cognitive model for lung cancer based on elements of evidence by Piao BingkuiA complex network association analysis algorithm was applied to construct a complex network association of the "disease-pattern element" relationship of lung cancer cases treated by Piao Bingkui.Eigenvector Centrality,average path length and Clustering Coefficient of nodes are used to evaluate the network and node properties.Finally,the core evidence elements and the correlations among the evidence elements are shown in the diagnosis and treatment of lung cancer by Prof.Piao Bingkui in terms of disease identification,staging,typing and staging.2.3.3 Construction of a treatment model for lung cancer based on pattern treatment by Piao BingkuiBased on the VBA program,we constructed the "disease-pattern" relationship matrix of lung cancer cases treated by Professor Piao Bingkui,and applied RStudio 2022.02.2 software to construct the evolution model of lung cancer staging,typing,and staging.IBM SPSS Statistics 26.0 software was applied.The patterns was used as the independent variable,and the subcutaneous symptoms,tongue body,tongue coating,pulse conditions,pathological examination,treatment stage and western medical diagnosis were used as covariates.Then the prediction model of lung cancer evidence diagnosis with binary logistic regression analysis was applied.Odds ratio was used to evaluate the degree of influence of this covariate in the diagnosis of lung cancer symptoms.3 Results3.1 Distribution of case informationAfter rigorous case screening,a total of 1,325 clinical cases of lung cancer treated by Professor Piao Bingkui,a nationally renowned Chinese medicine practitioner,were included in this study,with 95.09%of the patients’ age distribution between 40 and 70 and a male to female ratio of approximately 1:0.79.In the distribution of Western medical diagnostic information,21 ICD-11 diagnoses were involved,and more than 5%had lymph node metastatic malignancy,lung metastatic malignancy,bone or bone marrow metastatic malignancy,brain metastatic malignancy.and pleural effusion;AJCC stage Ⅰ(25.51%),stage Ⅲ(26.42%),stage Ⅱ(26.11%).and stage Ⅳ(21.96%)were relatively Evenly distributed;among the diagnoses involving pathology,non-small cell carcinoma predominated(86.19%),including adenocarcinoma of the lung(49.51%),and squamous cell carcinoma(18.34%).The stage of treatment involved post-operative(71.25%),post-chemotherapy(56.23%),and post-radiotherapy(20.75%).The distribution of TCM diagnoses involved 15 TCM disease names,of which malignant nucleus(21.58%),osteoma(8.60%),syncope(5.81%)and hanging drink(5.66%)accounted for a higher proportion;37 TCM symptoms were involved,with Qi deficiency of spleen and lung pattern(64.38%),phlegm-dampness accumulation(43.62%)and damp-heat-poison accumulation(36.60%)being the most widely distributed;36 TCM evidence elements were involved Of these,15 were pathogenic and 21 were pathogenic in nature,with the Spleen(46.79%),Liver(29.96%),Large Intestine(21.96%),Fire-Heat(87.77%),Qi Deficiency(55.62%)and Phlegm(55.47%)showing a high frequency of distribution among the evidence elements.The distribution of prescription medication information involved 23 lung cancer formulas and 160 Chinese herbal medicines,with the commonly used prescriptions being Sha Shen Mai Dong Fang(77.89%),Jian Quan Yi Qi Fang(18.04%)and Yu Ping Feng San(16.08%);the commonly used medicines were Licorice(99.25%),Huang Qi(94.57%),Shan Zao(89.51%),Mai Ye(89.28%),Liu Shen Qu(89.13%),Tu Fu Ling(88.98%),Bei Sha Shen(78.79%),Chen Pi(76.53%).Radix Platycodon(75.92%),Prince’s Ginseng(70.57%),and Fried Atractylodes(70.11%).Most of the herbal flavours are in the range of 16 to 20,and most of them belong to the spleen,lung and stomach meridians,and are mostly warm and flat in nature.3.2 Construction of a cognitive model for lung cancer based on elements of evidence by Piao BingkuiThe complex network association analysis algorithm built into the Gephi software was used to construct the ICD-11 case name diagnoses of lung cancer treated by Piao Bingkui:"malignant tumour of lung","metastatic malignant tumour of lymph nodes","metastatic tumour of bone or bone marrow".malignant tumour"."metastatic malignant tumour to brain","pleural effusion".AJCC stage "stage Ⅰ","stage Ⅱ","stage Ⅲ","Stage Ⅳ"、histopathology "non-small cell carcinoma" "small cell carcinoma "The results showed that the complex network was constructed with a mean of 1,000,000 patients.The results showed that the average path lengths of the constructed complex networks were balanced and comparable among the networks.The core evidence elements of the complex network associations for ICD-11 diagnosed lung malignancies were lung,spleen,blood stasis,fire-heat,dampness,qi deficiency and phlegm;the core evidence elements of the complex network associations for lymph node metastatic malignancies were lung,phlegm,qi stagnation,blood stasis and fire-heat;the core evidence elements of the complex network associations for bone or bone marrow metastatic malignancies were liver,kidney,tendons,bones and joints.The core elements of the complex network society for metastatic malignancies of the brain are heart and soul,lung and fire-heat;the core elements of the complex network society for pleural effusion are lung,blood stasis,drinking,water stagnation,phlegm and qi stagnation.Stage Ⅰ complex network society core evidence of lung malignancy AJCC stage Ⅰ is lung;stage Ⅱ complex network society core evidence is lung,phlegm,qi stagnation,fire-heat;stage Ⅲ complex network society core evidence is lung,and the pathological evidence with higher centrality of feature vectors are:phlegm,qi deficiency,fire-heat;stage Ⅳcomplex network society core evidence is kidney,lung,stomach,mind-shen,phlegm,qi stagnation,fire-heat,dampness,qi deficiency.The histopathological non-small cell carcinoma complex network society core evidence elements are lung,spleen,phlegm,fire-heat,blood stasis and dampness;the small cell carcinoma complex network society core evidence elements are lung and blood stasis.The core evidence of the complex network society after the treatment stage was lung,spleen,qi deficiency,fire-heat,blood stasis,dampness and phlegm;the core evidence of the complex network society after chemotherapy was spleen,liver,kidney,lung,stomach,heart and mind,dampness,phlegm,qi stagnation,fire-heat and qi deficiency;the core evidence of the complex network society after radiotherapy was lung,qi stagnation and fire-heat.3.3 Construction of a treatment model for lung cancer based on pattern treatment by Piao Bingkui3.3.1 Professor Piao Bingkui’s model of lung cancer pattern evolution in the treatment of lung cancerThe "disease-pattern" relationship matrix was constructed for 1325 cases of lung cancer treated by Professor Piao Bingkui,and the pattern evolution model was constructed based on the AJCC stage,western medical name,histopathological diagnosis,and treatment stage.The results showed that the main symptoms of AJCC stage Ⅰ cases were Qi deficiency of spleen and lung pattern,Phlegm dampness accumulation pattern and Damp-heat-poison accumulation pattern,while the main symptoms of stage Ⅱ cases were Qi deficiency of spleen and lung pattern,Phlegm dampness accumulation pattern,and Damp-heat-poison accumulation pattern.The main symptoms of stage Ⅱ cases are Qi deficiency of spleen and lung pattern,Phlegm dampness accumulation pattern and Damp-heat-poison accumulation pattern;the main symptoms of stage Ⅳ cases are deficiency of spleen and lung qi,Phlegm dampness accumulation pattern and Damp-heat-poison accumulation;In the case of "metastatic malignant tumor of lymph nodes" in Western medicine,the main patterns are Qi deficiency of spleen and lung pattern,Phlegm dampness accumulation pattern,and Damp-heat-poison accumulation.The main patterns of "metastatic malignant neoplasm of bone or bone or bone marrow" are Deficiency of liver and kidney pattern,Phlegm dampness accumulation pattern,and Qi deficiency of spleen and lung pattern;The main patterns of "malignant tumor of brain metastasis" are Blood stasis obstructing the brain collaterals pattern,Qi deficiency of spleen and lung pattern,and Damp toxin pattern;The main symptoms of "pleural effusion" are Phlegm dampness accumulation pattern,Qi deficiency of spleen and lung pattern,and Chest fluid retention pattern;Histopathological diagnosis of the case of "non-small cell carcinoma",in which the main patterns were Qi deficiency of spleen and lung pattern,Phlegm dampness accumulation pattern,and Damp-heat-poison accumulation,and the case of "small cell carcinoma",in which the main patterns were s Qi deficiency of spleen and lung pattern,Phlegm dampness accumulation pattern,and Damp-heat-poison accumulation pattern;During the "post-operative" stage of treatment,the main patterns were Qi deficiency of spleen and lung pattern,Phlegm dampness accumulation pattern,and Damp-heatpoison accumulation;During the "post-chemotherapy" stage,the main patterns were Qi deficiency of spleen and lung pattern,Phlegm dampness accumulation pattern,and Damp-heat-poison accumulation;During the "post-radiotherapy" stage,the main patterns were Qi deficiency of spleen and lung pattern,Phlegm dampness accumulation pattern,and Damp-heat-poison accumulation;During the "targeted therapy" stage,the main patterns were Qi deficiency of spleen and lung pattern,Damp-heat-poison accumulation pattern,and Phlegm dampness accumulation pattern.3.3.2 Predictive model for pattern-based treatment of lung cancer by Professor Piao BingkuiThe patterns of lung cancer "Qi deficiency of spleen and lung pattern "," Phlegm dampness accumulation pattern " and "Damp-heat-poison accumulation pattern " were used as the dependent variables,and "J subcutaneous symptoms","K tongue texture","L tongue coating","M pulse","O pathological examination ","V treatment stage" and"W western medical diagnosis" were used as covariates in the binary logistic regression analysis,the results showed that the percentages of correct predictions for the three models of "Qi deficiency of spleen and lung pattern" "Phlegm dampness accumulation pattern" and "Damp-heat-poison accumulation pattern "were 80.8%,74.7%,and 77.5%,respectively.The presence of metastatic malignant tumor of ovary,slow pulse,short breath,and spontaneous sweating were found to be positively correlated with the diagnosis of "Qi deficiency of spleen and lung pattern " in lung cancer cases treated by Professor Piao Bingkui(P<0.05);ICD-11 diagnosis of "soft tissue liposarcoma of the limbs",symptoms of "cough,coughing sputum,white sputum,heat in the hands and feet,shoulder pain,back pain,breath-holding,chest tightness,chest pain",tongue coating of "dry moss,thin moss "The presence of "floating pulse" was positively correlated with the diagnosis of "Phlegm dampness accumulation pattern"(P<0.05);ICD-11 diagnosis of "renal malignancy",treatment stage "post-immunotherapy,targeted therapy,post-operative",symptoms "yellow phlegm,irritable heat in the five hearts,ulceration of the cheeks,bitterness in the mouth,coughing up blood,joint pain,dryness of the throat,spontaneous sweating".tongue color and tongue color.The presence of the elements of tongue color and texture of "dentate tongue" was positively correlated with the diagnosis of "Damp-heat-poison accumulation" in lung cancer(P<0.05).ICD-11 diagnosis of "renal malignancy",treatment stage "post-immunotherapy,targeted therapy,post-operative",symptoms "yellow phlegm,irritable heat in the five hearts,ulceration of the cheeks,bitterness in the mouth,coughing up blood.joint pain,dryness of the throat,spontaneous sweating",tongue color and tongue color,The presence of the elements of tongue color and texture of "dentate tongue" was positively correlated with the diagnosis of "Damp-heat-poison accumulation" in lung cancer(P<0.05).3.3.3Treatment based on pattern identification model of lung cancer by Professor Piao BingkuiUsing "Qi deficiency of spleen and lung pattern " as the basis for classification,and 445 disease,symptom,prescription and medicine elements to build a Bayesian network of "disease-symptom-evidence-prescription-medicine" for Qi deficiency of spleen and lung pattern,the model classification accuracy was 72.68%,and found that adenocarcinoma,post-operative,post-chemotherapy,cough,light red tongue and thin moss were the main basis for his diagnosis of "Qi deficiency of spleen and lung pattern".The prescription is mostly a self-prepared Shashen Qingfei Formula.The main drugs used are Gancao,Huangqi,Shanzha,Tufuling,Liushenqu,Maiya,Beishanshen,Jiegeng,Chenpi,Chaobaizhu,Beishashen,Shengyiyiren.Using "Phlegm dampness accumulation pattern" as the basis for classification,a Bayesian network of "disease-disease-evidence-prescription-medicine" was constructed with 445 disease,symptom,prescription and medicine elements to classify"phlegm and dampness with nodules",and Correctly Classified Instances was 70.19%.It was found that Professor Piao Bingkui used post-operative,postchemotherapy,cough,coughing,light red tongue,thin moss,and white moss as the main diagnostic basis for the diagnosis of "Phlegm dampness accumulation pattern".The main prescription used was the self-prepared Shashen Qingfei Formula,with the following main drugs:Gancao,Huangqi,Shanzha,Tufulling,Maiya,Liushenqu,Beishashen,Chenpi,Jiegeng,Taizishen,Chaobaizhu,Shengyiyiren.Using "Damp-heat-poison accumulation pattern" as the basis for classification,we constructed a Bayesian network of 445 disease,symptom,prescription and medicine elements for "disease-symptom-evidence-prescription-medicine" of Damp-heat-poison accumulation pattern,and the Correctly Classified Instances was 64.38%.It was found that adenocarcinoma,postoperative,post-chemotherapy,pale red tongue,thin coating were the main basis for Professor Piao Bingkui to diagnose "Damp-heat-poison accumulation pattern",and the main prescriptions used were Self-prepared Shashen Qingfei Formula,Jianpi Yiqi Formula,Yu Ping Feng San,and the main drugs used were Gancao,Huangqi,Shanzha,Maiya,shanzha,Liushenqu,Tufulling,Beishashen,Jiegeng,Chenpi,Chaobaizhu,Taizishen,Shengyiyiren.Using "Deficiency of liver and kidney pattern" as the classification basis,a Bayesian network of "disease-disease-evidence-prescription-medicine" was constructed for 445 disease,symptom,prescription and medicine elements of 1325 cases of lung cancer treated by Prof.Piao.Correctly Classified Instances was 87.09%.After chemotherapy,a light red tongue with thin coating was found to be the main basis for Professor Piao Bingkui diagnosis of "Deficiency of liver and kidney pattern".The main prescription is Self-prepared Shashen Qingfei Formula,the main drugs are Gancao,Huangqi,Tufulling,Shanzha,Liushenqu,Maiya,Baizhu,Taizishen,Chenpi,Beishashen,Jiegeng,Shengyiyiren.Based on the classification of "Blood stasis obstructing the brain collaterals pattern",a Bayesian network of "disease-symptom-evidence-prescription-medicine"was constructed for 445 disease,symptom,prescription and medicine elements of 1325 cases of lung cancer treated by Professor Park Byung-kyu.Correctly Classified Instances was 95.47%.It was found that metastatic malignant tumor,stage Ⅳa,adenocarcinoma,post-chemotherapy,post-radiotherapy,light red tongue and thin moss were the main basis for Professor Piao’s diagnosis of "Blood stasis obstructing the brain collaterals pattern".The main prescription is:Changpu Yujin Formula,the main herbs are Gancao,Tufuling,Huangqi,Shanzha,Liushenqu,Maiya,Chenpi,Shengyiyiren,Tiazishen,Shichangpu,Yujin,Baizhu,Yizhi.4 ConclusionIn order to explore the theoretical system of Professor Piao Bingkui’s treatment based on pattern differentiation of lung cancer,this study included 1325 outpatient cases of lung cancer treated by Prof.Piao Bingkui,and applied frequency statistics,complex network association discovery algorithm,logistic regression analysis,and Bayesian network algorithm to construct Professor Park’s patterncognitive model.lung cancer evidence prediction model,and lung cancer pattern treatment model.Based on the comparison between the models,the study concluded that the unique treatment based on pattern differentiation system of lung cancer by Professor Piao Bingkui has the following characteristics.Firstly,he advocated the integration of Chinese and Western medicine,and emphasized the importance of Western medical information on diagnosis,AJCC stage and treatment stage in the theoretical system of lung cancer diagnosis and treatment.Secondly,he advocated that "combined deficiency and excess" is the pathogenesis of lung cancer,and that "deficiency of the spleen and lung pattern","damp phlegm pattern" and "damp heat and toxin pattern" as the basic patterns.Thirdly.he advocates the treatment concept of "reinforce healthy qi to strengthen the body".In the treatment of lung cancer,drugs with strengthen the spleen and lung are used as the basic drugs,such as Gancao,Huangqi,Tufuling,Liushenqu,Maiya,Beishashen,Jiegeng,Chenpi,Baizhu,Taizishen,Yiyiren.If there is deficiency of the liver and kidney pattern,add Gouqizi,Yizhi,Nvzhenzi,Weilingxian,etc.If there is blood stasis obstructing the brain collaterals pattern,add Shi changpu and Yujin for regulate qi and transform phlegm,open the orifices and lead the medicine upwards,add Jiangcan and Quanxie enhancing the power to move and open the channels. |
