Objective To explore the value of machine learning method(LASSO regression)in constructing a model for predicting intracranial aneurysm rupture based on CT angiography(CTA).Methods Clinical data and CTA examination results of 133 patients with intracranial aneurysms were collected,and they were divided into a ruptured group(103 cases)and an unruptured group(30 cases)based on whether they were ruptured.Compare the differences in clinical data and CTA examination parameters between the two groups of patients;Use Logistic regression and LASSO regression to screen for risk factors related to aneurysm rupture,and then construct a predictive model.Evaluate the predictive value of the model using the receiver operating characteristic curve(ROC)and area under the curve(AUC).Results Compared with the patients in the unruptured group,the patients in the ruptured group had more cases of diabetes history,more cases of irregular tumor shape,more cases of combined cysts,and higher incidence angles.Single factor logistic regression showed that the history of diabetes,tumor shape and number of sacs,and incidence angle were related to aneurysm rupture.The model built with these four indicators had a moderate predictive effect on aneurysm rupture,with an AUC value of 0.766.LASSO regression screening showed that the history of diabetes,tumor shape,number,width,incidence angle and number of sacs were significantly related to the rupture of intracranial aneurysms.The model constructed had a high prediction efficiency,with an AUC value of 0.902.Conclusion The history of diabetes,the shape of aneurysm,the presence of ascus,and the angle of incidence are related to the rupture of aneurysm.The prediction model constructed by LASSO regression can better predict the risk of intracranial aneurysm rupture.
维普
万方数据