Application Of Machine Learning In The Prediction Of Osteoporosis In Ovariectommzed Rats

Postmenopausal osteoporosis is a common disease associated with aging in humans.Because of estrogen deficiency bone structure changed and bone mass became lower.Also can increases bone fragility and is prone to fracture then causes pain and death.The internal environment of bone growth is related to the level of biochemical indicators of bone metabolism and mechanical stimulation,many scholars have done a lot of research on these two aspects.Based on the theory of machine learning and bone remodeling,this paper proposes five biochemical indicators related to bone metabolism in postmenopausal osteoporosis based on the pathogenesis of osteoporosis.The model of female ovariectomized rat can correctly simulate low estrogen osteoporosis.Based on the rat femur trabecular finite element model and the bone remodeling theory,the initial bone remodeling threshold was obtained.Then,the rate of bone mineral density in the diabetic disease group and the normal group in the same period was obtained.And the initial bone remodeling threshold was corrected one by one.The machine learning model Random Forest,Adaboost,GBDT are built after the data is cleaned by the machine learning theory.The rate of change of the processed biochemical indicators is taken as the characteristics of the machine learning training,and the corresponding bone rebuilding threshold is taken as a label and brought into the established machine learning model for training.The trained model is used to predict the bone remodeling threshold and compared with the standard bone remodeling threshold.Then compared training effect between different training models,and selected the most accurate one to further optimized.And the model is evaluated after optimization to verify the accuracy of the model.This paper establishes a quantitative relationship between postmenopausal osteoporosis biochemical markers of bone metabolism and bone remodeling thresholds through a machine learning model.The results show that the machine learning model can predict the bone remodeling threshold well through biochemical indicators.In the three models in this paper,the random forest method has the best prediction effect.This article links the biochemical indicators from the mechanical point of view to explain the cause of osteoporosis,and provides an available method for the prevention and treatment of osteoporosis.