Research Of Health Status Identification Algorithm Based On TCM Theory Of State

"The State of Traditional Chinese Medicine" is the foundation of the theory of traditional Chinese medicine treatment of disease,providing theoretical guidance and top-level design for treatment of disease and health management.State and health status identification is an important component of TCM ethology.The scientificity and accuracy of algorithm models directly affect the development of health status identification.The establishment of intelligent and accurate identification algorithm model is the key problem of health status identification.This research studies the algorithm model of TCM health status identification from theory,clinical data sets and artificial intelligence methods.ObjectiveConstruct a TCM state identification model algorithm to provide technical support for the assessment of TCM health status.On the one hand,it can improve the maturity of TCM health status recognition algorithms for clinical application;on the other hand,exploring model algorithms based on machine learning in the context of big data helps to quickly optimize the algorithm model itself in order to adapt to complex and changeable Application scenarios.Method1.Using literature analysis,focus on the concept and connotation of state and health status identification,research on health status identification mode,research on health status identification algorithm,and existing problems and solutions.2.Utilize clinical data collected from 1146 cases to construct a TCM health status data set,conduct research on TCM health status identification model algorithms,and use classic artificial intelligence algorithms K nearest neighbor(kNN),support vector machine(SVM),and decision tree(DT),And artificial neural network(BP-MLL)to build the algorithm model,and compare its performance.3.Adopt new multi-label classification algorithms ML-kNN,RankSVM,LIFT to optimize the TCM health status identification model,and analyze the performance characteristics of multi-label and single-label classification algorithms in TCM health status identification.Result1.The system has sorted out the conceptual connotation of states and health states,and discussed the health state identification mode from "artificial identification mode" and "intelligent identification mode";discussed the algorithm from the state identification process of "characterization parameters-state elements-state names" The matching problem between the model and the health status identification is that the health status identification is essentially a classification problem,and the classification algorithm can be used to construct the health status identification algorithm model.2.In the comparison experiment between single-label classification algorithm and neural network BP-MLL algorithm: average accuracy: BP-MLL 82.31%,SVM 67.24%,DT 45.64%,kNN 30.94%.The experimental results show that the BP-MLL algorithm has higher average accuracy than the single-label classification algorithms kNN,SVM,and DT.At the same time,in the single-label classification algorithm comparison experiment,the SVM classification performance is significantly better than the kNN and DT performance.3.In the optimization research of multi-label classification algorithm: average accuracy:RankSVM 86.98%,LIFT 85.38%,ML-kNN 68.63%,BP-MLL 64.69%,SVM 64.53%,DT47.02%,kNN 33.41% In the comparison between multi-label classification algorithm and single-label classification algorithm: ML-kNN has better classification effect than kNN,RankSVM and LIFT algorithms have better classification effect than SVM.The classification effect of RankSVM and LIFT algorithm in multi-label classification is higher than that of BP-MLL and ML-kNN.RankSVM has the best classification effect.Conclusion1.The use of artificial intelligence technology machine learning methods(including single-label classification,multi-label classification)can build an algorithm model of TCM health status recognition;2.Multi-label classification method has certain significance for the establishment and optimization of TCM health state recognition algorithm model,and plays a certain role in solving the mixed state problems;3.Multi-label classification algorithm In the construction and optimization of Chinese medicine health state identification algorithm model,the performance of RankSVM algorithmis optimal.