Research On The Analysis And Transmission Of Professor Piao Bingkui’s Characteristic Experience In The Treatment Of Lung Cancer Based On The Combination Of Disease And Pattern Differentiation

1 ObjectiveBased on various machine learning algorithms,a large number of Professor Piao Bingkui’s lung cancer cases were systematically analyzed from multiple perspectives.Based on the personalized treatment model combining disease and evidence,the individualized cognitive model of famous veteran Chinese medicine practitioners was constructed.In addition,we will summarize Professor Piao’s clinical experience and academic thoughts on the treatment of lung cancer.2 Method2.1 Summarizing and summarizing the experience of human useBased on the compilation of Professor Piao’s medical cases,medical discourse,experience collections,and oral presentations,we summarize and conclude Professor Piao’s diagnosis and treatment of lung cancer in terms of disease identification,evidence identification,treatment rules,prescriptions,medication rules,and flexible treatment methods.His unique academic thinking is also analyzed.2.2 Construction of machine learning cognitive models2.2.1 Data collection and entryThe relevant medical case data from Prof.Piao’s clinic were collected and organized,and these data were used as the study subjects,and 1026 cases that met the criteria were screened according to the inclusion and exclusion criteria.A dataset of lung cancer treatment by Prof.Piao was established,and finally,according to the unified criteria,two people entered the case data individually into Microsoft Excel for Mac(16.58)for storage and exported it for backup as baseline information.2.2.2 Data processing and quality controlThe data in the dataset are categorized according to the unified processing specification,including noun disaggregation and noun unification.The uniform model case collection is also applied,and two people enter the data separately to strictly control the data quality.With the help of the Excel built-in VBA program,the conversion of Boolean data expressions is completed.2.2.3 Descriptive statistics and machine learningThe statistical results such as frequency,probability,and dispersion of baseline demographics of the case,chief complaint,duration of illness,Western medicine diagnosis,Chinese medicine diagnosis,symptoms,treatment,major prescriptions,and medication were described using Excel software.Then the core data of disease,evidence,method,prescription,and medicine were analyzed by Bayesian networks,support vector machines,complex network visualization,systematic clustering,association rules,and other algorithms using Weka(Waikato Environment for Knowledge Analysis),Gephi(0.9.2 201709242018),SPSS Statistics 26.0,R 4.1.2 and RStudio 2022.02.1.Finally,a cognitive model of the combination of disease and evidence in the diagnosis and treatment of lung cancer by Professor Piao Bingkui was constructed.3 Conclusion3.1 Results of descriptive statisticsAfter a strict screening of inclusion and exclusion criteria,a total of 1026 cases meeting the criteria were included,of which 641 cases(62.5%)were male patients and 385 cases(37.5%)were female patients;the age distribution mainly covered the population aged 40-80 years,with an average age of 61.69 ± 12.20 years.Among the 357 first lung cancer cases included in the analysis,a total of 32 symptoms,including thirst and excessive drinking,were observed.The three most frequent symptoms were cough in 168 cases(47.06%),irritating dry cough in 47 cases(13.17%),and postoperative lung malignancy in 46 cases(12.89%).Among the 1026 lung cancer cases included in the analysis,a total of 13 Western diagnosed diseases were present.In addition to lung cancer,the top three were bronchial or lung adenocarcinoma in 593 cases(57.80%),lymph node metastatic malignancies in 398 cases(38.79%),and bronchial or lung squamous cell carcinoma in 143 cases(13.94%);a total of 32 TCM symptoms were present,with the top three being spleen and lung deficiency in 734 cases(71.54%),phlegm obstructing lung in 493 cases(22.81%),and damp-heat 22.81%),and 442 cases(42.01%)of dampness-heat and toxicity;a total of 33 TCM treatments,including tonifying the spleen and benefiting the lung(734 cases,71.54%),clearing heat and dampness(430 cases,41.91%),and invigorating blood and detoxifying toxins(430 cases,41.91%)were the most common treatments;a total of 42 main formulas appeared,with the top three being Sha Shen Mai Dong Formula 769 cases(accounting for 74.95%),Gua Lou Xie Bai Ban Xia Soup 138 cases(accounting for 13.45%)and Yu Ping Feng Particles 108 cases(accounting for 10.53%);a total of 149 drugs such as Glycyrrhizae Radix Et Rhizoma and Astragali Radix appeared,of which 1016 cases(accounting for 99.03%)were Glycyrrhizae Radix Et Rhizoma,977 cases(accounting for 95.22%)were Astragali Radix,Crataegi Fructus,Massa Medicata Fermentata and Hordei Fructus Germinatus accounted for 91.72%,91.62%and 91.52%respectively,and 919 cases of Smilacis Glabrae Rhizoma(89.57%)were the most commonly used drugs by Professor Piao.3.2 Intrinsic relationship of cognitive modelsWith the help of the one-way weighted network of western medical disease names and symptoms of lung cancer established by Gephi 0.9.2,the one-way association of western medical disease names and Chinese medical symptoms was performed in a complex network.The relationship between Western medical disease names(lymph node metastatic malignancy,bone metastatic malignancy,liver metastatic malignancy,pleural effusion,pericardial effusion)and symptoms of tumor metastasis was taken as the main object of study,and a total of 116 groups of corresponding Western medical diagnoses(metastatic sites and complications)and symptom relationships were obtained.The pathological types with weights of 20 and above and the corresponding symptoms were extracted,and 26 groups of correlations were obtained.It was found that there were clear correlations between lymph node metastasis and phlegm blocking lung syndrome,bone metastasis,and liver and kidney deficiency syndrome,liver metastasis and dampness-heat and toxicity embedding syndrome,brain metastasis and stasis blocking brain ligament syndrome,pleural effusion and drink stopping chest syndrome,and pericardial effusion and drink stopping pericardium syndrome.On the contrary,in the complex network relationship between pathological staging and symptoms,there was no obvious consistency and uniformity in the symptoms of adenocarcinoma,squamous carcinoma,small cell carcinoma,and large cell carcinoma.The relationship between pathological staging and treatment was explored through the Bayesian network,and "squamous carcinoma","adenocarcinoma","small cell carcinoma",and "large cell carcinoma" were used as classification attributes."The results showed that there was no clear correlation between pathological staging and treatment method in Professor Piao’s treatment philosophy and that pathological staging was not a major consideration in his clinical practice.This result was complemented by Gephi’s complex network visualization.Based on the Bayesian network to explore the relationship between metastasis and treatment method,1026 cases of lung malignant tumors treated by famous veteran Chinese medicine practitioner Piao Byung-kyu were analyzed with 38 attributes as"lymph node metastatic malignant tumor","brain metastatic malignant tumor","bone metastatic malignant tumor","liver metastatic malignant tumor",and "pleural metastatic malignant tumor"."bone metastatic malignant tumor","liver metastatic malignant tumor","pleural metastatic malignant tumor" were used as classification attributes for Bayesian network analysis,and 165 structural relationships between disease(evidence)and method were found.The classification accuracy was 65.59%,97.76%,93.86%,97.75%and 97.47%,and the kappa values were 0.2202,0.7179,0.4811,-0.0019 and 0.1205,respectively.From this,165 relationships between tumor metastasis-treatment were found,and the treatment will change and differ accordingly depending on the tumor metastasis.For example,the common treatments for brain metastatic malignant tumors are Transform stasis and unblock collaterals,Clear heat and eliminate dampness,Clear heat and remove toxins and Circulate blood and transform stasis,Open the orifices with aroma,Tonify qi and blood;the common treatments for bone metastatic malignant tumors are Tonify the spleen and benefit the lung,Transform stasis and unblock collaterals,Clear heat and remove toxins and Circulate blood and transform stasis,Clear heat and eliminate dampness,Open the orifices with aroma;the common treatments for liver metastatic malignant tumors are Tonify the spleen and benefit the lung,Clear heat and eliminate dampness,Clear heat and remove toxins and Circulate blood and transform stasis,Regulate qi and resolve masses,Transform phlegm and dissipate nodules,Disperse dampness,Transform phlegm and dissipate nodules;common treatments for pleural metastatic malignant tumors are Tonify the spleen and benefit the lung,Clear heat and eliminate dampness,Clear heat and remove toxins and Circulate blood and transform stasis,Regulate qi and resolve masses,Transform phlegm and dissipate nodules,Disperse dampness,Transform phlegm and dissipate nodules.Clear heat and remove toxins Circulate blood and transform stasis,Regulate qi and resolve masses,Transform phlegm and dissipate nodules.On this basis,we constructed a complex network visualization diagram with and without distant metastases,which shows that the treatment without distant metastases is relatively concentrated,while with distant metastases shows a relatively wide range of treatments,with respective treatment tendencies depending on the invasion site of metastases.Based on the disease-square relationship model of the support vector machine,the classification of 1026 cases of pulmonary malignant tumors treated by Piao Byung-kyu,a famous Chinese medicine practitioner,with 44 attributes of "liver metastatic malignant tumor","bone metastatic malignant tumor","lymph node metastatic malignant tumor" and "brain metastatic malignant tumor" as the classification attributes were analyzed by SMO based on support vector machine.The accuracy of SMO analysis based on the support vector machine was 97.856%,93.957%,65.302,and 98.246%,respectively.We also derived the weighting relationships between "disease-fang",in which the higher absolute values for bone metastatic malignant tumors were:Gu Zhuan Yi Formula=YES(-2.001),Chang Pu Yu Jin Soup=YES(-2.000);the higher absolute values for lymph node metastatic malignant tumors were:Lin Ba Jie Formula=YES(2.000),Si Jun Zi Soup=YES(-2.000),Xiao Chai Hu Soup=YES(-1.999).The higher absolute values of weights were:Lin Ba Jie Formula=YES(-2.000),Si Jun Zi Soup=YES(-2.000),and Xiao Chai Hu Soup=YES(-1.999);and the higher absolute values of weights corresponding to malignant tumors with brain metastases were:Chang Pu Yu Jin Soup=YES(1.999).Based on the Support Vector Machine(SVM)"disease(evidence)-drug"relationship model,we analyzed 1026 cases of pulmonary malignant tumors treated by Prof.Piao with 151 attributes as "liver metastatic malignant tumor","bone metastatic malignant tumor","lymph node metastatic malignant tumor" and "brain metastatic malignant tumor"."The accuracy of the classification was 97.86%,93.18%,64.23%,and 97.76%,respectively.64.23%and 97.76%,respectively.The weight relationship between "disease(evidence)-drug" was obtained,among which the absolute value of the weight corresponding to the liver metastatic malignant tumor is higher:Phragmitis Rhizoma=YES(1.737),Dipsaci Radix=YES(0.862),Achyranthis Bidentatae Radix=YES(0.826);the absolute value of the weight corresponding to the bone metastatic malignant tumor is higher:Phragmitis Rhizoma=YES(-1.734),Dynariae Rhizoma=YES(-1.334),Acori Tatarinowii=YES(-1.081),Achyranthis Bidentatae Radix=YES(-1.017);the absolute value of the weight corresponding to lymph node metastatic malignant tumor is higher:Anemarrhenae Rhizoma=YES(-1.502),Cocias Semen=YES(-1.637),Stemonae Radix=YES(-1.680),Asparagi Radix=YES(-2.095);the absolute value of the weight corresponding to brain metastatic malignant tumor is higher:Asparagi Radix=YES(-2.095),Stemonae Radix=YES(-1.680),Cocias Semen=YES(1.637).Using SPSS Statistics 26.0,based on the systematic clustering method for the frequency of occurrence ≧ 100 times of traditional Chinese medicine,using Pearson correlation,combined with the clinical experience of Professor Piao Bingkui,coclustering out of the pairing 9 groups.Including Portfolio 1:Crataegi Fructus,Medicated Leaven,Hordei Fructus Germinatus,Citri Reticulatae Pericarpium,Poria,Amomi Fructus Rotundus;Portfolio 2:Dioscoreae Rhizoma,Aurantii Fructus,Atractylodis Macrocephalae Rhizoma,Alpiniae Oxyphyllae Fructus;Portfolio 3:Smilacis Glabrae Rhizoma,Cocias Semen,Lycii Fructus,Astragali Radix,Pseudostellariae Radix,Glycyrrhizae Radix Et Rhizoma,Bombyx Batryticatus;Portfolio 4:Curcuma Rhizoma;Portfolio 5:Cinnamomi Cortex,Angelicae Sinensis Radix,Jujubae Fructus,Paeoniae Radix Rubra,Paridis Rhizoma,Schisandrae Chinensis Fructus,Belamcandae Rhizoma,Polygoni Cuspidati Rhizoma Et Radix,Solanum Nigrum;Portfolio 6:Solanum Lyratum,Scutellariae Barbatae Herba;Portfolio 7:Ligustri Lucidi Fructus,Atractylodis Macrocephalae Rhizoma,Glehniae Radix,Platycodonis Radix,Ophiopogonis Radix,Armeniacae Semen Amarum,Agrimoniae Herba,Hedyotidis Diffusae Herba;Portfolio 8:Trichosanthis Fructus,Allii Macrostemonis Bulbus,Pinelliae Rhizoma;Portfolio 9:Corni Fructus,Saposhnikoviae Radix,Fagopyri Dibotryis Rhizoma.The core drug group discovery was based on association rules,30116 association rules were established for 1026 instances and 141 attributes of medical cases of lung malignant tumors diagnosed and treated by famous veteran Chinese medicine practitioner Piao Bingkui.Based on the Apriori association rule frequent itemset algorithm found that Zingiberis Rhizoma Recent-Jujubae Fructus,Trichosanthis Fructus-Allii Macrostemonis Bulbus embodies an extremely strong association in single-drug association and Zingiberis Rhizoma Recent-Smilacis Glabrae RhizomaJujubae Fructus,Astragali Radix-Zingiberis Rhizoma Recent-Jujubae Fructus,Pinelliae Rhizoma-Allii Macrostemonis Bulbus-Trichosanthis Fructus,Pseudostellariae Radix-Allii Macrostemonis Bulbus-Trichosanthis Fructus,Citri Reticulatae Pericarpium-Allii Macrostemonis Bulbus-Trichosanthis Fructus,Platycodonis Radix-Allii Macrostemonis Bulbus-Trichosanthis Fructus,Smilacis Glabrae Rhizoma-Allii Macrostemonis Bulbus-Trichosanthis Fructus,Astragali RadixAllii Macrostemonis Bulbus-Trichosanthis Fructus,often in a combined form.4 ConclusionThis study combines supervised learning with unsupervised learning,human experience with machine learning,and qualitative research with quantitative research based on the construction of a disease-evidence combined lung cancer diagnosis and treatment model.The experience of Professor Piao in the diagnosis and treatment of lung cancer is constructed as a cognitive model,and the experience of Professor Piao in the diagnosis and treatment of lung cancer is mined.We hope to complement and inform the excavation of the ideas of famous veteran TCM practitioners and their academic schools...