Study On Early Warning Model Of Coronary Heart Disease Adverse Events Based On Inquiry And Pulse Diagnosis Information Of Traditional Chinese Medicine

Objective:The objective of this study is to optimize the objective feature extraction method of pulse diagnosis and its inclusion index according to the physiological and pathological characteristics of patients with coronary heart disease,analyze the high-risk syndrome types and the characteristics of inquiry and pulse diagnosis of coronary heart disease adverse events,and establish the early warning model of coronary heart disease adverse events based on TCM inquiry and pulse diagnosis information,to provide an economic and convenient supplementary methods for early warning of coronary heart disease adverse events,and provide a certain scientific basis for the popularization and application of objective technology of four diagnosis of traditional Chinese medicine.Methods:472 Patients diagnosed with coronary heart disease from January 2016 to December 2019 in Shanghai Hospital of traditional Chinese medicine and other medical institutions were followed up for the longest period of 5 years,with cerebral infarction,heart failure,myocardial infarction,renal failure or cardiovascular death as the end point of follow-up.In order to improve the accuracy and anti-interference ability of the pulse feature extraction algorithm,a single period pulse waveform quality evaluation model is established by using a variety of machine learning algorithms.Combined with the previous research results of the research group,a series of pulse parameters such as moving average parameters and pulse rate variability parameters were extracted and included in the research scope.On this basis,Cox proportional hazard model and random survival forest model were used to analyze the relationship between traditional Chinese medicine inquiry/pulse parameters/syndrome types and the incidence of various adverse events,and a comprehensive early warning model of coronary heart disease adverse events with all adverse events as the end events was established based on inquiry and pulse information.The accuracy of the model was evaluated by the time-dependent receiver operating curve,and the test set was divided into high-risk group and low-risk group according to the maximum threshold of Youden index on the training set.Kaplan-Meier survival curve and log-rank test were used to determine whether there was significant difference in survival between the two groups.Results:（1）The logistic regression model with x₁,x₃,x₄ and x₁₄ features can complete the task of single period pulse waveform quality evaluation in most cases,and its area under the ROC curve on the test set can reach 0.9920.（2）The indexes included in the random survival forest model are mainly pulse parameters.The proportion and importance ranking of the new pulse parameters such as D_e/n₁ and t_0.8are higher than those of the traditional time-domain parameters.（3）Heart kidney yin deficiency syndrome,heart blood stasis syndrome,phlegm obstructing heart vessels syndrome and qi stagnation and blood stasis syndrome are the high risk syndrome types of secondary cerebral infarction in patients with coronary heart disease.Qi stagnation and blood stasis syndrome,phlegm obstructing heart vessels syndrome and heart blood stasis syndrome are the high risk syndrome types of secondary myocardial infarction in patients with coronary heart disease.Yang qi deficiency and heart blood stasis are high risk syndromes of secondary heart failure in patients with coronary heart disease.Phlegm obstructing heart vessels syndrome and yang qi deficiency syndrome are the high risk syndrome types of secondary renal failure in patients with coronary heart disease.In patients with coronary heart disease,the risk of secondary adverse events in patients with qi deficiency and blood stasis syndrome and qi yin deficiency syndrome is relatively low,while the risk of secondary adverse events in patients with qi stagnation and blood stasis syndrome,phlegm obstructing heart vessels syndrome,heart blood stasis syndrome,heart kidney yin deficiency syndrome and yang qi deficiency and failure syndrome is relatively high.（4）Chest pain,shortness of breath,preference for meat and headache are common risk factors of secondary adverse events in patients with coronary heart disease.Pulse parameters such as D_e/n₁,t₄,t₁,h₃/h₁,t_0.8/t₁ and h_f/4 play an important role in the early warning model of coronary heart disease adverse events,and show significant non-linear effect.（5）Based on the information of traditional Chinese medicine inquiry and pulse diagnosis,the comprehensive early warning model of coronary heart disease adverse events established by Lasso-Cox proportional hazard model only includes one pulse diagnosis parameter.The area under the td ROC curve of the model in the test set for 2 years and 3years is 0.7178 and 0.6382 respectively.According to the model,the hazard ratio of high-risk group to low-risk group is 4.820.The top 30 indicators in the comprehensive early warning model of coronary heart disease adverse events established by the random survival forest model include 12 pulse diagnosis parameters.The area under the td ROC curve of the model in the test set for 2 years and 3 years is 0.8072 and 0.8139 respectively.The hazard ratio of the high-risk group to the low-risk group is 6.067,The area under the td ROC curve and the risk ratio are higher than those of Lasso-Cox proportional hazards model.Conclusion:（1）The quality evaluation model of single cycle pulse waveform based on four input features and logistic regression algorithm is helpful to improve the accuracy and anti-interference ability of pulse diagnosis information extraction algorithm.（2）New pulse indexes such as D_e/n₁ and t_0.8 can supplement part of the missing information of traditional time-domain characteristic pulse parameters to a certain extent,and play an important role in the early warning model of coronary heart disease adverse events.（3）The risk degree of secondary adverse events in patients with coronary heart disease is lower in patients with qi deficiency and blood stasis syndrome and qi yin deficiency syndrome,but higher in patients with qi stagnation and blood stasis syndrome,phlegm obstructing the heart vessels syndrome,heart blood stasis syndrome,heart kidney yin deficiency syndrome and yang qi deficiency and decline syndrome.This group of traditional Chinese medicine syndromes can reflect the evolution trend of coronary heart disease.（4）Inquiry parameters such as chest pain,shortness of breath,preference for meat,headache and pulse parameters such as D_e/n₁,t₄,t₁,h₃/h₁,t_0.8/t₁,h_f/4 provide a good guidance for judging the condition and evaluating the prognosis of patients with coronary heart disease.（5）The comprehensive early warning model of coronary heart disease adverse events based on Cox proportional hazard model and random survival forest model can effectively assess the risk of various secondary adverse events in patients with coronary heart disease.The random survival forest model can make full use of inquiry and pulse information because it can analyze the nonlinear effect of indicators,to achieve higher recognition accuracy and higher hazard ratio.