| Objective:Primary liver cancer（PLC）,referred to as liver cancer for short,is one of the most common malignant tumors that seriously threaten human health.Traditional Chinese medicine can obviously improve the coexistence of multiple symptoms in patients with advanced liver cancer,relieve their discomfort and improve their quality of life.This study uses the real-world electronic medical records of TCM liver cancer,studies the classification of liver cancer symptom groups based on the hidden structure model,explores the related influencing factors of the occurrence and development of liver cancer symptom groups,and retrospectively analyzes and summarizes the drug-symptom relationship and drug compatibility characteristics of TCM in treating liver cancer with various data mining methods,which is of great significance for clinicians to improve liver cancer patients’ symptom groups more pertinently and take timely and effective intervention measures.Methods:1.A total of 565 structured TCM electronic medical records clinically diagnosed as "primary liver cancer" were collected from April 2017 to May2019 in Hubei Provincial Hospital of Traditional Chinese Medicine.With the help of "human-computer collaborative phenotypic spectrum annotation system"（www.tcmai.org）,the text of the medical records were structured,entity-labeled,and manually reviewed,and the key information such as positive symptoms,diseases,Chinese medicine,and prognosis were extracted in batches,classified into databases and standardized processing was performed.2.Combined with the 131 standardized TCM symptoms,the TCM symptom hidden structure model of liver cancer was constructed using Lantern 5.0 software.Combined with the professional theoretical knowledge of traditional Chinese medicine,the model learning interpretation and comprehensive clustering were performed on the classification results to extract the classification results of symptom groups that were clinically significant and in line with the characteristics of pathogenesis of liver cancer.The similarity between that symptom of patients and symptom group is calculated,and the patients and the symptom groups are match to obtain a liver cancer patient subgroup divided by the symptom groups.3.Based on the length of hospital stay and prognosis information of patients,kaplan-Meier was used to draw the survival curve,and log-rank test was used to evaluate the difference in survival between symptom groups,and the difference in prognosis based on symptom groups was analyzed to obtain the typical liver cancer symptom groups with predictive significance for the prognosis of liver cancer.Combined with patients’ baseline information,intervention measures and clinical outcome（survival or death）and other key information,multiple regression analysis was used to analyze the occurrence,development and prognostic factors of symptom clusters.4.Combined with symptom and TCM prescription database,frequency statistics were used to analyze the four qi,five flavor and meridian of drugs.With the help of SPSS 25.0 and SPSS Modeler 14.1 analysis software,complex network analysis,cluster analysis and association rule analysis were carried out for liver cancer symptom and prescription respectively.Explore the core drug compatibility and drug-syndrome distribution characteristics of liver cancer.Result:A total of 131 TCM symptoms（i.e.,significant variables）were included in the study to construct the hidden structure model of liver cancer symptoms,and 14 hidden variables were obtained.Through model interpretation and comprehensive clustering,six related TCM syndrome types were summed up,which were liver stagnation and spleen deficiency syndrome,yang deficiency and water flooding syndrome,damp heat toxin accumulation syndrome,liver and kidney yin deficiency syndrome,liver stagnation and qi stagnation syndrome（liver and stomach incompatibility syndrome）,and phlegm stagnation and blood stasis syndrome.On account of this,the six symptom clusters of liver cancer were named after Cluster1-6,which specifically reflected in the types of mental state related symptom cluster,digestive system related symptom cluster,cancer fatigue related symptom cluster,respiratory cycle related symptom cluster,etc.Through symptom similarity analysis,six liver cancer populations divided by symptom clusters were obtained,in which Cluster1,Cluster3,and Cluster4 were the symptom clusters in the high proportion of patient populations.Combined with the baseline information of the patient and interventions,Logistic regression was performed and found that the relevant factors affecting the symptom clusters of liver cancer included age,gender,drinking history,hepatitis history,as well as surgical treatment and TCM intervention.Study content two included 420 cases of liver cancer with TCM prescriptions,and a total of 114 independent symptoms and 1,622 TCM prescriptions were extracted.The results of complex network analysis showed that the symptoms of fatigue,poor appetite,poor sleep,liver discomfort,abdominal distension,dry mouth and yellow urine were strongly linked.The results showed that the drug properties of liver cancer drugs were mainly cold and warm,and the drug taste was mainly Ganping drug,followed by bitter drug.The meridian tropism of drugs was concentrated in the spleen and lung meridians,followed by the liver and stomach meridians.The high-frequency drugs were selected and analyzed by complex network and association rules,and the results showed that there was a strong link between drugs such as Rhizoma Atractylodis Macrocephalae,Poria,Radix Astragali seu Hedysari,Radix Salviae Miltiorrhizae,Herba Artemisiae Scopariae,Herba Scutellariae Barbatae,and Rhizoma Pinelliae.The common drug pairs for liver cancer were: stir-fried Fructus Hordei Germinatus-stir-fried Fructus oryzae Germinatus Poria-herba hedyotidis diffusae,radix astragali;Poria-herba hedyotidis diffusae,and rhizoma atractylodis macrocephalae;White atractylodes rhizome,radix pseudostellariae and poria;Poria-Massa Medicata Fermentata,and Radix Salviae Miltiorrhizae.Finally,high-frequency drugs were selected for clustering analysis,and four drug clusters were obtained.The first type:（1）stir-fried rice sprouts and stir-fried malt;（2）Fructus Crataegi,Massa Medicata Fermentata,and Endothelium Corneum Gigeriae Galli;（3）Rhizoma Dioscoreae and Coicis Semen;（4）Carapax Trionycis,Concha Ostreae,Radix Curcumae,and Fructus Gleditsiae Abnormalis.The second type:（1）Radix Bupleuri,Fructus Aurantii,Radix Angelicae Sinensis,and Radix Codonopsis;（2）Radix Ophiopogonis,Radix Rehmanniae and Cortex Moutan.The third type:（1）Radix Paeoniae Alba and Rhizoma Corydalis;（2）Magnolia officinalis,Fructus Aurantii Immaturus,Rhizoma Pinelliae,and Pericarpium Citri Reticulatae;（3）Fructus Trichosanthis and Radix Scutellariae.The fourth type:（1）Rhizoma Alismatis,Polyporus,Herba plantaginis,and Herba Artemisiae Scopariae;（2）Herba Hedyotis and Herba Scutellariae Barbatae;（3）Rhizoma Atractylodis Macrocephalae,Poria,Radix Glycyrrhizae,and Radix Pseudostellariae;（4）Rhizoma Bletillae and Rhizoma Coptidis;（5）Radix Salviae Miltiorrhizae,Radix Astragali,and Fructus Forsythiae.Conclusion:The research method of mining the symptom groups of liver cancer and dividing the liver cancer population by establishing the hidden structure model is explored,and the research results show that there are differences in influencing factors causing different symptom groups,which provides a new research idea for the research of symptom group classification and the research of dividing the disease population based on the symptom groups extracted based on the basic theory of traditional chinese medicine.Combined with a variety of data mining methods,it was summarized that liver cancer was mainly characterized by deficiency-excess syndrome such as fatigue,poor appetite,poor sleep,liver discomfort,abdominal distension,dry mouth,and little urine,and it involved the elements of "toxin","phlegm","blood stasis",and "deficiency".The relationship between drugs and symptoms generally met the requirements of early treatment,based on strengthening the body resistance and invigorating the spleen,and symptomatic treatment in the late stage,with the resolving phlegm,removing blood stasis,soothing the liver,softening the hard mass,detoxification and inhibiting cancer as the solutions. |
PDF Full Text Request