Study On Auxiliary Diagnosis And Treatment Based On Artificial Neural Network And Association Mining Technology

With the development of social and economic,people pay more attention to their own health problems than before,and the demand for medical and health services in China is also growing.However,the distribution of medical resources is imbalanced not only between east and west,but also between urban and rural areas.Patients tend to go to superior hospitals for treatment and it leads to the lack of experience in primary hospitals.The level of primary medical care decreases,and the willingness of patients to seek medical treatment in primary hospitals decreases,which leads to a vicious circle.The development of information technology makes the application of electronic medical records universal,which produces a large number of electronic medical data,and also provides the possibility for artificial intelligence to extract experience.If we can introduce the diagnosis and treatment experience of the superior hospital into the primary hospital and improve the medical level of them,the difficulties above will be effectively alleviated.According to the characteristics of diagnosis part and treatment part of medical experience,the classification algorithm and the association rule mining algorithm are used for extracting.There are many kinds of classification algorithm and association rule mining algorithm,so we need to choose the appropriate algorithm to form the diagnosis and treatment experience system.In order to improve the effect of experience extraction,it is necessary to improve the algorithm.At the same time,in order to enhance the effect of experience extraction,it is necessary to improve the algorithm.The main research work of this paper is as follows:(1)The main algorithm of diagnosis experience extraction is determined.As candidate algorithms,BP neural network and support vector machine have good universality and accuracy,but their own parameters are difficult to determine and will significantly affect their performance.Aiming at the problem of BP neural network and support vector machine algorithm,this paper uses swarm intelligence algorithm for optimizing.By searching for the minimum value of the test functions,it is confirmed that the optimization accuracy of Grey Wolf algorithm is significantly higher than that of particle swarm optimization algorithm and genetic algorithm.In order to verify the practical application effect,experiments were carried out on Hungarian data set,and grey wolf BP neural network was compared with other diagnostic experience extraction candidate algorithms combined with different swarm intelligence algorithms and different classification algorithms.The experiment shows that the accuracy of the support vector machine models combined with genetic algorithm,particle swarm optimization algorithm and grey wolf algorithm are improved in turn,which shows that grey wolf algorithm also has the best optimization ability in practical application.The accuracy of grey wolf BP neural network algorithm is higher than that of all support vector machine models,which shows that BP neural network has better ability of extracting diagnosis experience than support vector machine.Grey wolf BP neural network algorithm is initially used as diagnosis experience extraction algorithm.(2)The original grey wolf algorithm has shortcomings.In this paper,it is improved as IGWO algorithm to enhance the optimization ability.Then the IGWO and BP neural network algorithm are combined to form IGWO-BP neural network algorithm in order to improve the performance of diagnosis experience extraction.To solve the problem of uneven initial population of grey wolf algorithm,tent chaotic map is used for generating initial population.The convergence factor of the original algorithm declines linearly,which cannot fully meet the requirements of global search and local search ability in the search process.In this paper,a new nonlinear convergence factor is introduced to improve the algorithm.In the later stage of the algorithm,it is easy to fall into the local optimum,so the leader-wolf-disturbance mechanism is introduced.The disturbance probability is dynamically adjusted with the iteration process to help wolves jump out of the local optimum.Then the improved grey wolf algorithm is compared with the improved swarm intelligence algorithm in other literatures by the optimizations of single-peak functions and multi-peak functions.The results show that the improved grey wolf algorithm has the best optimization performance compared with other improved algorithms.This algorithm is combined with BP neural network algorithm to form IGWO-BP neural network algorithm.The IGWO-BP neural network is compared with other BP neural networks combined with improved swarm intelligence algorithm by experiments carried out on UCI data sets.The results show that the accuracy of IGWO-BP neural network algorithm is further improved than GWO-BP neural network,and higher than BP neural networks combined with other improved swarm intelligence algorithms.IGWO-BP neural network algorithm is decided as the final algorithm for diagnosis experience extraction.(3)Aiming at the problem that the performance of the original FP-growth algorithm is significantly reduced by repeatedly accessing the elements,this paper introduces tail tag domain and odd tag field to FP growth algorithm and it is improved as TOFP-growth algorithm.Tail tag field and odd tag field make the skip access in the header table and among the FP tree nodes possible,then the total number of visits is reduced and efficiency is improved.The data set of Frequent Itemset Mining Dataset Repository was used for the experiment.The results show that when the amount of data is small,the effect of tail tag is similar to that of odd tag.When the amount of data is large,the effect of odd tag is weaker,and tail tag is still able to improve the performance significantly.The TOFP-growth algorithm which combines the two methods has good performance under the above conditions.It shows that the TOFP-growth is significantly more efficient than the original algorithm and has good robustness.Finally,this paper experiments on the medical data set and the common drug combinations corresponding to hepatobiliary diseases are mined out.TOFP-growth algorithm and IGWO-BP neural network algorithm constitute a complete diagnosis and treatment experience extraction system.