Risk Factors Of Infantile Hemangioma Based On Data Mining

Infantile hemangioma is one of the most common tumors in infants and young children.The disease can cause eye capillaries to dilate and cause permanent scars.It can also affect vision and cause other more serious problems,such as disfigurement or even death.The occurrence of infantile hemangioma diseases is related to a variety of risk factors.How to detect and effectively prevent and control these risk factors as early as possible is of great significance to control the occurrence of infantile hemangioma diseases.In this paper,the actual case data of Liuzhou Workers’ Hospital is used as a training sample,and whether infants and young children suffer from hemangioma is regarded as a binary classification problem to study the risk factors related to infantile hemangioma.Logistic regression algorithm and CART decision tree algorithm are used to establish the risk factor model of infantile hemangioma,the performance of the two models was compared,and satisfactory results were obtained.Among them,the classification accuracy rate based on the Logistic regression model on the test set is85.83%,and the model evaluation index,the area under the ROC curve is 0.90;the classification accuracy rate based on the decision tree model on the test set is 86.67%,AUC is 0.93,where the CART decision tree model performs better on the experimental data set.At the same time,the risk factors obtained based on the two classification models were analyzed.The results of logistic regression model showed that sex,weight,premature delivery,family history of hemangioma,use of fetal-care drugs,birth age,viral infection during pregnancy,abnormal placenta,and excessive vitamin use were the 9 main risk factors for infantile hemangioma.;The results of CART decision tree model based on grid search optimization showed that gender,family history of hemangioma,premature delivery,first birth,use of fetal care drugs,physical labor,harmful substances,breastfeeding,smoking,spontaneous delivery were the 10 main risk factors for infantile hemangioma.Among them,gender,family history of hemangioma,premature delivery,and taking fetal protection drugs are the main risk factors jointly selected by the two models.According to the above analysis results,the model of infantile hemangioma risk factors based on CART decision tree algorithm has better performance and has certain guiding effect and practical application value for the prevention of infantile hemangioma.