Application Of BP Neural Network In Abnormal Bone Metabolism Of Type Ⅰ Diabetes

With the improvement of living standards,the prevalence of type 1 diabetes is increasing year by year.The long-term hyperglycemia in patients with type 1 diabetes causes bone metabolism disorder,which leads to type 1 diabetic osteoporosis.Osteoporosis results in the reduction of bone quality and bone mass,which leads to fracture and even disability and death in diabetic patients.According to the theory of bone remodeling,bone has functional adaptability and is regulated by the mechanical environment and internal environment.The internal environment is related to biochemical indicators of bone metabolism,and the mechanical environment is related to mechanical stimulation.Many scholars have done a lot of research on these two aspects,but there are few quantitative studies combining the two.This paper mainly constructs the mathematical relationship between mechanical and biochemical indicators,predicts the changes of bone mineral density through the changes of biochemical indicators,prevents diseases in advance,and assists the treatment of diabetic osteoporosis patients.Based on the theory of bone remodeling and neural network theory,this paper proposes 17 biochemical indicators related to bone metabolism in type 1 diabetes according to the mechanism of diabetes treatment.The concentration changes of 17 biochemical indices in the diabetic group and the normal group were established through the experimental data of SD(Sprague Dawley)rats.The biochemical indexes of the diabetic group and the normal group were worse,and the difference was recorded.According to the theory of bone remodeling,three different finite element models of trabecular bone were selected in the right femur of normal SD rats,and different material properties were assigned to three trabecular beam models to calculate the initial apparent density of different models,compared with the bone density of normal SD rats,the appropriate finite element model of trabecular bone was selected.On the basis of the finite element model,the finite element bone reconstruction procedure was programmed to determine the initial bone remodeling threshold in the bone remodeling equation.The threshold of initial bone remodeling is revised one by one,according to the change rate of bone mineral density in the same period of diabetes mellitus group compared with normal group.and the corrected bone reconstruction threshold and the data corresponding to the difference of the biochemical indicators are arranged.BP(Back Propagation)neural network model is constructed by neural network theory.The difference of biochemical indexes is taken as input data,and the corresponding bone remodeling threshold is taken as output data.They was substituted into BP neural network for training,and the BP neural network model is used to predict the bone remodeling threshold.At the same time,it is compared with the multiple linear regression model.The predicted threshold of bone remodeling was substituted into the finite element program of bone remodeling to predict the change of bone mineral density,and the accuracy of the BP neural network model was verified by comparing with the actual situation.In this paper,the quantitative relationship between biochemical indexes of bone metabolism and bone remodeling threshold in type 1 diabetes mellitus was established by BP neural network model,which is helpful to prevent osteoporosis in type 1 diabetes mellitus and provide reference for the treatment of osteoporosis.