Research On The Diagnostic Model Of Idiopathic Pulmonary Fibrosis With Qi-yin Deficiency And Blood Stasis Syndrome Based On Data Mining

Objective: This study uses related symptom information collected by the clinic on the syndromes of idiopathic pulmonary fibrosis cases to explore the idea of combining traditional Chinese medicine theories with cutting-edge science and technology,to construct a diagnostic model that is suitable for idiopathic pulmonary fibrosis with deficiency of both qi and yin and blood stasis Syndrome.This diagnostic model can reduce the risk of failure treatment and mistreatment to the disease in the clinic,and provide a more objective basis and more convenient means for the clinical diagnosis of the disease.Method: Two deputy senior experts independently entered the relevant case information of patients with idiopathic pulmonary fibrosis in the inpatient and outpatient departments of the Respiratory Department of the Affiliated Hospital of Liaoning University of Traditional Chinese Medicine from September 2018 to November 2020 in Excel.They also independently diagnosed relevant cases based on diagnostic criteria and personal experience.If the dialectical results of these two experts are inconsistent or cause disagreement,it will be resolved by the two parties through consultation or be handed over to an authoritative expert(the chief physician)for judgment.After the data information is preprocessed,the decision tree,neural network,support vector machine,and other algorithms in SPSS 24.0 and SPSS Modeler 18.0 are independently used to establish diagnosis models of idiopathic pulmonary fibrosis with deficiency of both qi and yin and blood stasis.By comparing the discrimination accuracy,sensitivity,specificity,and availability of the training set(75% of the total sample)and the test set(25% of the total sample)of each diagnostic model,calculating the area under the ROC curve of the neural network model,each constructed diagnostic model would be assessed and evaluated to filtrate the best diagnostic model and important variables that can indirectly reflect its clinical application value.Result:1.To input 348 related cases to establish an idiopathic pulmonary fibrosis database,it included 44 syndrome indicators and excluded 11 indicators with a frequency of less than10%.Finally,33 syndrome indicators were used to construct a diagnostic model.2.To establish a decision tree model with CHAID,QUEST,CRT,C5.0,multilayer perceptron neural network model,support vector machine model,the availability of these models is greater than 0.5.The neural network model test set has the highest accuracy(92.63%),and the C5.0 decision tree model has the highest sensitivity(100%),and the support vector machine model has the highest specificity(95.45%).However,the sensitivity and specificity of the neural network model are relatively high(both greater than 90%).The area under the ROC curve is 0.988.The model evaluation is dominated by discrimination accuracy.Considering all factors,the neural network model is the best diagnosis model.Conclusion:1.The IPF diagnosis model of deficiency of both qi and yin and blood stasis syndrome based on neural network is the best diagnosis model.2.The dark tongue is the best distinguishing feature of the diagnostic model,and it has the largest contribution rate.Then it should be combined with symptoms such as chest tightness,dry mouth,bruising lips,abnormal sweating,little phlegm,petechiae on the tongue,sluggishness,and other symptoms to assist in the diagnosis of IPF Qi and Yin deficiency with blood stasis syndrome.