Research On The Prediction Of Clinical Path Based On Improved LSTM

Most of the traditional clinical pathways in hospitals are made by the medical staff using their own experience with strong subjectivity.Because of the complexity of clinical data,strong privacy,too many patients and large individual differences,the treatment plan of patients within a single disease is different,so it is difficult to establish a typical clinical path for each disease.According to the data characteristics of clinical pathway,this paper combines the LSTM structure of deep learning recurrent neural network with the activity prediction theory of process mining,and applies it to the prediction of medical clinical pathway.The clinical pathway contains sequential diagnosis and treatment activities,and our method is to use LSTM neural network to predict the next activity in the clinical pathway.In this paper,the basic model structure of LSTM neural network is improved by introducing convolution layer and pooling layer of convolution neural network,so a clinical path prediction model based on improved LSTM is proposed.The research of this paper has a certain practical significance for the application of deep learning,process mining and other technologies in the prediction of clinical pathway activities in medical field.For medical staff,it can help them to establish the clinical pathway of disease better,improve the efficiency of medical treatment,and reduce the waste of time and resources.In this paper,the real data of medical scene is used to evaluate the performance of the model,and compared with other prediction models to verify the effect of the proposed model.The experimental results show that the clinical pathway prediction model based on the improved LSTM,compared with the basic CNN and LSTM model,has achieved good results in accuracy,cross entropy loss and complexity.