Research On Clinical Pathway Optimization Of Type 2 Diabetes Based On Deep Learning

In recent years,with the arrival of the new coronavirus epidemic,the national government and the broad masses of the people have started to pay attention to the development of the medical field,and the clinical pathway has become the focus of nationwide medical reform.With the rise of big data technology and artificial intelligence technology,the problems of low resource utilization and a high mutation rate in traditional clinical pathway schemes have gradually emerged.Depending on the latest published data,the number of patients with type 2 diabetes in China has exceeded 170 million,and the problem of low resource utilization and high variation rate of type 2diabetes clinical pathways is more prominent.Therefore,it is critical to optimize the clinical pathway of type 2 diabetes.In order to solve the above problems,this paper proposes the following two methods to optimize the clinical pathway of type 2 diabetes:(1)Optimize the resource scheduling of clinical pathways to improve the utilization of medical resources.Firstly,according to the sequence relationship between doctor’s order tasks in the clinical pathway of type 2 diabetes mellitus and the quantity relationship of therapeutic resources required by each doctor’s order task,this paper constructed an optimization model of clinical pathway resource scheduling.Secondly,the multi-objective particle swarm optimization algorithm was utilized to find out the clinical path resource scheduling scheme with the shortest treatment days and the largest number of patients.At the same time,Wolf individual position updating method in the Gray Wolf algorithm is integrated into the multi-objective particle swarm optimization algorithm to avoid the situation that the optimal resource scheduling scheme cannot be found.Finally,an example of type 2 diabetes clinical pathways was taken to verify the effectiveness of the algorithm and solve the problem of clinical pathway resource scheduling optimization.(2)To predict patients’ blood glucose levels to reduce the rate of clinical pathway variation.Firstly,the collected patient characteristics,such as blood sugar value and carbohydrate intake in the past 2 hours,were reprocessed and used as input data for the multi-task LSTM prediction model.Secondly,this paper uses multi-task learning technology to improve the LSTM prediction model to simultaneously predict the blood glucose values of patients in the next 30 minutes,60 minutes,90 minutes and 120 minutes.Finally,as there are many factors affecting the predictive value of blood glucose in patients,hospitals generally do not conduct comprehensive monitoring on patients during treatment to collect all the factors affecting blood glucose.Therefore,this paper further finds out the key characteristics of patients with a greater impact on the predictive value of blood glucose in patients.It is found that the difference between the root-mean-square error obtained by using the patient’s key characteristic information as input data and the root-mean-square error obtained by using all the patient’s characteristic information as input data is less than 0.5.It is proved that it is feasible to use key patient characteristics to predict blood glucose value,and it is more conducive to the application in the treatment of type 2 diabetes clinical pathways,saving hospital resource consumption.In this paper,the deep learning technology is introduced into the study of optimizing clinical pathways,and the clinical pathways for type 2 diabetes are optimized from the aspects of resource scheduling and variation rate,which provides the direction for the future optimization of clinical pathways and guarantees for the utilization of medical resources and the control of medical costs.At the same time,it is easy to introduce additive attention mechanism and improve the prediction ability of the model.