This study identified factors influencing the dislodgement of CT-guided hookwire localization. Univariate analysis showed that PS interval was the only significant factor for dislodgement. We utilized various multivariate analysis and machine learning techniques such as SMOTE and random forest for deeper analysis. In both the random forest model and XGBoost model, five factors—DNP, DWP, total depth, and age—showed relatively high priority in predicting dislodgement.
